Machine Learning
See recent articles
Showing new listings for Tuesday, 3 June 2025
- [1] arXiv:2506.00030 [pdf, html, other]
-
Title: Modality Equilibrium Matters: Minor-Modality-Aware Adaptive Alternating for Cross-Modal Memory EnhancementComments: work in progressSubjects: Machine Learning (cs.LG)
Multimodal fusion is susceptible to modality imbalance, where dominant modalities overshadow weak ones, easily leading to biased learning and suboptimal fusion, especially for incomplete modality conditions. To address this problem, we propose a Shapley-guided alternating training framework that adaptively prioritizes minor modalities to balance and thus enhance the fusion. Our method leverages Shapley Value-based scheduling to improve the training sequence adaptively, ensuring that under-optimized modalities receive sufficient learning. Additionally, we introduce the memory module to refine and inherit modality-specific representations with a cross-modal mapping mechanism to align features at both the feature and sample levels. To further validate the adaptability of the proposed approach, the encoder module empirically adopts both conventional and LLM-based backbones. With building up a novel multimodal equilibrium metric, namely, equilibrium deviation metric (EDM), we evaluate the performance in both balance and accuracy across four multimodal benchmark datasets, where our method achieves state-of-the-art (SOTA) results. Meanwhile, robustness analysis under missing modalities highlights its strong generalization capabilities. Accordingly, our findings reveal the untapped potential of alternating training, demonstrating that strategic modality prioritization fundamentally balances and promotes multimodal learning, offering a new paradigm for optimizing multimodal training dynamics.
- [2] arXiv:2506.00039 [pdf, html, other]
-
Title: AbsoluteNet: A Deep Learning Neural Network to Classify Cerebral Hemodynamic Responses of Auditory ProcessingSubjects: Machine Learning (cs.LG); Sound (cs.SD); Audio and Speech Processing (eess.AS)
In recent years, deep learning (DL) approaches have demonstrated promising results in decoding hemodynamic responses captured by functional near-infrared spectroscopy (fNIRS), particularly in the context of brain-computer interface (BCI) applications. This work introduces AbsoluteNet, a novel deep learning architecture designed to classify auditory event-related responses recorded using fNIRS. The proposed network is built upon principles of spatio-temporal convolution and customized activation functions. Our model was compared against several models, namely fNIRSNET, MDNN, DeepConvNet, and ShallowConvNet. The results showed that AbsoluteNet outperforms existing models, reaching 87.0% accuracy, 84.8% sensitivity, and 89.2% specificity in binary classification, surpassing fNIRSNET, the second-best model, by 3.8% in accuracy. These findings underscore the effectiveness of our proposed deep learning model in decoding hemodynamic responses related to auditory processing and highlight the importance of spatio-temporal feature aggregation and customized activation functions to better fit fNIRS dynamics.
- [3] arXiv:2506.00131 [pdf, html, other]
-
Title: Adapting Offline Reinforcement Learning with Online DelaysSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Offline-to-online deployment of reinforcement-learning (RL) agents must bridge two gaps: (1) the sim-to-real gap, where real systems add latency and other imperfections not present in simulation, and (2) the interaction gap, where policies trained purely offline face out-of-distribution states during online execution because gathering new interaction data is costly or risky. Agents therefore have to generalize from static, delay-free datasets to dynamic, delay-prone environments. Standard offline RL learns from delay-free logs yet must act under delays that break the Markov assumption and hurt performance. We introduce DT-CORL (Delay-Transformer belief policy Constrained Offline RL), an offline-RL framework built to cope with delayed dynamics at deployment. DT-CORL (i) produces delay-robust actions with a transformer-based belief predictor even though it never sees delayed observations during training, and (ii) is markedly more sample-efficient than naïve history-augmentation baselines. Experiments on D4RL benchmarks with several delay settings show that DT-CORL consistently outperforms both history-augmentation and vanilla belief-based methods, narrowing the sim-to-real latency gap while preserving data efficiency.
- [4] arXiv:2506.00135 [pdf, html, other]
-
Title: Tradeoffs between Mistakes and ERM Oracle Calls in Online and Transductive Online LearningSubjects: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS); Machine Learning (stat.ML)
We study online and transductive online learning when the learner interacts with the concept class only via Empirical Risk Minimization (ERM) or weak consistency oracles on arbitrary instance subsets. This contrasts with standard online models, where the learner knows the entire class. The ERM oracle returns a hypothesis minimizing loss on a given subset, while the weak consistency oracle returns a binary signal indicating whether the subset is realizable by some concept. The learner is evaluated by the number of mistakes and oracle calls. In the standard online setting with ERM access, we prove tight lower bounds in both realizable and agnostic cases: $\Omega(2^{d_{VC}})$ mistakes and $\Omega(\sqrt{T 2^{d_{LD}}})$ regret, where $T$ is the number of timesteps and $d_{LD}$ is the Littlestone dimension. We further show that existing online learning results with ERM access carry over to the weak consistency setting, incurring an additional $O(T)$ in oracle calls. We then consider the transductive online model, where the instance sequence is known but labels are revealed sequentially. For general Littlestone classes, we show that optimal realizable and agnostic mistake bounds can be achieved using $O(T^{d_{VC}+1})$ weak consistency oracle calls. On the negative side, we show that limiting the learner to $\Omega(T)$ weak consistency queries is necessary for transductive online learnability, and that restricting the learner to $\Omega(T)$ ERM queries is necessary to avoid exponential dependence on the Littlestone dimension. Finally, for certain concept classes, we reduce oracle calls via randomized algorithms while maintaining similar mistake bounds. In particular, for Thresholds on an unknown ordering, $O(\log T)$ ERM queries suffice; for $k$-Intervals, $O(T^3 2^{2k})$ weak consistency queries suffice.
- [5] arXiv:2506.00136 [pdf, html, other]
-
Title: On Designing Diffusion Autoencoders for Efficient Generation and Representation LearningComments: 21 pages, 10 tables, 15 figuresSubjects: Machine Learning (cs.LG)
Diffusion autoencoders (DAs) are variants of diffusion generative models that use an input-dependent latent variable to capture representations alongside the diffusion process. These representations, to varying extents, can be used for tasks such as downstream classification, controllable generation, and interpolation. However, the generative performance of DAs relies heavily on how well the latent variables can be modelled and subsequently sampled from. Better generative modelling is also the primary goal of another class of diffusion models -- those that learn their forward (noising) process. While effective at adjusting the noise process in an input-dependent manner, they must satisfy additional constraints derived from the terminal conditions of the diffusion process. Here, we draw a connection between these two classes of models and show that certain design decisions (latent variable choice, conditioning method, etc.) in the DA framework -- leading to a model we term DMZ -- allow us to obtain the best of both worlds: effective representations as evaluated on downstream tasks, including domain transfer, as well as more efficient modelling and generation with fewer denoising steps compared to standard DMs.
- [6] arXiv:2506.00152 [pdf, html, other]
-
Title: Aligning Language Models with Observational Data: Opportunities and Risks from a Causal PerspectiveComments: 10+12 pages, 8 figuresSubjects: Machine Learning (cs.LG); Econometrics (econ.EM); Machine Learning (stat.ML)
Large language models are being widely used across industries to generate content that contributes directly to key performance metrics, such as conversion rates. Pretrained models, however, often fall short when it comes to aligning with human preferences or optimizing for business objectives. As a result, fine-tuning with good-quality labeled data is essential to guide models to generate content that achieves better results. Controlled experiments, like A/B tests, can provide such data, but they are often expensive and come with significant engineering and logistical challenges. Meanwhile, companies have access to a vast amount of historical (observational) data that remains underutilized. In this work, we study the challenges and opportunities of fine-tuning LLMs using observational data. We show that while observational outcomes can provide valuable supervision, directly fine-tuning models on such data can lead them to learn spurious correlations. We present empirical evidence of this issue using various real-world datasets and propose DeconfoundLM, a method that explicitly removes the effect of known confounders from reward signals. Using simulation experiments, we demonstrate that DeconfoundLM improves the recovery of causal relationships and mitigates failure modes found in fine-tuning methods that ignore or naively incorporate confounding variables. Our findings highlight that while observational data presents risks, with the right causal corrections, it can be a powerful source of signal for LLM alignment. Please refer to the project page for code and related resources.
- [7] arXiv:2506.00158 [pdf, html, other]
-
Title: Privacy Amplification in Differentially Private Zeroth-Order Optimization with Hidden StatesSubjects: Machine Learning (cs.LG)
Zeroth-order optimization has emerged as a promising approach for fine-tuning large language models on domain-specific data, particularly under differential privacy (DP) and memory constraints. While first-order methods have been extensively studied from a privacy perspective, the privacy analysis and algorithmic design for zeroth-order methods remain significantly underexplored. A critical open question concerns hidden-state DP analysis: although convergent privacy bounds are known for first-order methods, it has remained unclear whether similar guarantees can be established for zeroth-order methods. In this work, we provide an affirmative answer by proving a convergent DP bound for zeroth-order optimization. Our analysis generalizes the celebrated privacy amplification-by-iteration framework to the setting of smooth loss functions in zeroth-order optimization. Furthermore, it induces better DP zeroth-order algorithmic designs that are previously unknown to the literature.
- [8] arXiv:2506.00166 [pdf, html, other]
-
Title: Disentangled Safety Adapters Enable Efficient Guardrails and Flexible Inference-Time AlignmentComments: 16 pages, 2 figures, including references and appendixSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Existing paradigms for ensuring AI safety, such as guardrail models and alignment training, often compromise either inference efficiency or development flexibility. We introduce Disentangled Safety Adapters (DSA), a novel framework addressing these challenges by decoupling safety-specific computations from a task-optimized base model. DSA utilizes lightweight adapters that leverage the base model's internal representations, enabling diverse and flexible safety functionalities with minimal impact on inference cost. Empirically, DSA-based safety guardrails substantially outperform comparably sized standalone models, notably improving hallucination detection (0.88 vs. 0.61 AUC on Summedits) and also excelling at classifying hate speech (0.98 vs. 0.92 on ToxiGen) and unsafe model inputs and responses (0.93 vs. 0.90 on AEGIS2.0 & BeaverTails). Furthermore, DSA-based safety alignment allows dynamic, inference-time adjustment of alignment strength and a fine-grained trade-off between instruction following performance and model safety. Importantly, combining the DSA safety guardrail with DSA safety alignment facilitates context-dependent alignment strength, boosting safety on StrongReject by 93% while maintaining 98% performance on MTBench -- a total reduction in alignment tax of 8 percentage points compared to standard safety alignment fine-tuning. Overall, DSA presents a promising path towards more modular, efficient, and adaptable AI safety and alignment.
- [9] arXiv:2506.00172 [pdf, html, other]
-
Title: Breakpoint: Scalable evaluation of system-level reasoning in LLM code agentsComments: 21 pages, 14 figuresSubjects: Machine Learning (cs.LG)
Benchmarks for large language models (LLMs) have predominantly assessed short-horizon, localized reasoning. Existing long-horizon suites (e.g. SWE-bench) rely on manually curated issues, so expanding or tuning difficulty demands expensive human effort and evaluations quickly saturate. However, many real-world tasks, such as software engineering or scientific research, require agents to rapidly comprehend and manipulate novel, complex structures dynamically; evaluating these capabilities requires the ability to construct large and varied sets of problems for agents to solve. We introduce Breakpoint, a benchmarking methodology that automatically generates code-repair tasks by adversarially corrupting functions within real-world software repositories. Breakpoint systematically controls task difficulty along two clear dimensions: local reasoning (characterized by code complexity metrics such as cyclomatic complexity) and system-level reasoning (characterized by call-graph centrality and the number of simultaneously corrupted interdependent functions). In experiments across more than 900 generated tasks we demonstrate that our methodology can scale to arbitrary difficulty, with state-of-the-art models' success rates ranging from 55% on the easiest tasks down to 0% on the hardest.
- [10] arXiv:2506.00175 [pdf, html, other]
-
Title: Accountability Attribution: Tracing Model Behavior to Training ProcessesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Modern AI development pipelines often involve multiple stages-pretraining, fine-tuning rounds, and subsequent adaptation or alignment-with numerous model update steps within each stage. This raises a critical question of accountability: when a deployed model succeeds or fails, which stage is responsible, and to what extent? We pose the problem of accountability attribution, which aims to trace model behavior back to specific stages of the training process. To address this, we propose a general framework that answers counterfactual questions about stage effects: how would the model behavior have changed if the updates from a training stage had not been executed?. Within this framework, we introduce estimators based on first-order approximations that efficiently quantify the stage effects without retraining. Our estimators account for both the training data and key aspects of optimization dynamics, including learning rate schedules, momentum, and weight decay. Empirically, we demonstrate that our approach identifies training stages accountable for specific behaviors, offering a practical tool for model analysis and a step toward more accountable AI development.
- [11] arXiv:2506.00181 [pdf, html, other]
-
Title: On the Interaction of Noise, Compression Role, and Adaptivity under $(L_0, L_1)$-Smoothness: An SDE-based ApproachComments: This manuscript is a work in progress: We welcome commentsSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Using stochastic differential equation (SDE) approximations, we study the dynamics of Distributed SGD, Distributed Compressed SGD, and Distributed SignSGD under $(L_0,L_1)$-smoothness and flexible noise assumptions. Our analysis provides insights -- which we validate through simulation -- into the intricate interactions between batch noise, stochastic gradient compression, and adaptivity in this modern theoretical setup. For instance, we show that \textit{adaptive} methods such as Distributed SignSGD can successfully converge under standard assumptions on the learning rate scheduler, even under heavy-tailed noise. On the contrary, Distributed (Compressed) SGD with pre-scheduled decaying learning rate fails to achieve convergence, unless such a schedule also accounts for an inverse dependency on the gradient norm -- de facto falling back into an adaptive method.
- [12] arXiv:2506.00188 [pdf, html, other]
-
Title: Cluster-Aware Causal Mixer for Online Anomaly Detection in Multivariate Time SeriesSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Early and accurate detection of anomalies in time series data is critical, given the significant risks associated with false or missed detections. While MLP-based mixer models have shown promise in time series analysis, they lack a causality mechanism to preserve temporal dependencies inherent in the system. Moreover, real-world multivariate time series often contain numerous channels with diverse inter-channel correlations. A single embedding mechanism for all channels does not effectively capture these complex relationships. To address these challenges, we propose a novel cluster-aware causal mixer to effectively detect anomalies in multivariate time series. Our model groups channels into clusters based on their correlations, with each cluster processed through a dedicated embedding layer. In addition, we introduce a causal mixer in our model, which mixes the information while maintaining causality. Furthermore, we present an anomaly detection framework that accumulates the anomaly evidence over time to prevent false positives due to nominal outliers. Our proposed model operates in an online fashion, making it suitable for real-time time-series anomaly detection tasks. Experimental evaluations across six public benchmark datasets demonstrate that our model consistently achieves superior F1 scores.
- [13] arXiv:2506.00198 [pdf, html, other]
-
Title: MOFGPT: Generative Design of Metal-Organic Frameworks using Language ModelsSrivathsan Badrinarayanan, Rishikesh Magar, Akshay Antony, Radheesh Sharma Meda, Amir Barati FarimaniComments: 21 pages, 3 figures (in main text, without references)Subjects: Machine Learning (cs.LG); Materials Science (cond-mat.mtrl-sci); Artificial Intelligence (cs.AI)
The discovery of Metal-Organic Frameworks (MOFs) with application-specific properties remains a central challenge in materials chemistry, owing to the immense size and complexity of their structural design space. Conventional computational screening techniques such as molecular simulations and density functional theory (DFT), while accurate, are computationally prohibitive at scale. Machine learning offers an exciting alternative by leveraging data-driven approaches to accelerate materials discovery. The complexity of MOFs, with their extended periodic structures and diverse topologies, creates both opportunities and challenges for generative modeling approaches. To address these challenges, we present a reinforcement learning-enhanced, transformer-based framework for the de novo design of MOFs. Central to our approach is MOFid, a chemically-informed string representation encoding both connectivity and topology, enabling scalable generative modeling. Our pipeline comprises three components: (1) a generative GPT model trained on MOFid sequences, (2) MOFormer, a transformer-based property predictor, and (3) a reinforcement learning (RL) module that optimizes generated candidates via property-guided reward functions. By integrating property feedback into sequence generation, our method drives the model toward synthesizable, topologically valid MOFs with desired functional attributes. This work demonstrates the potential of large language models, when coupled with reinforcement learning, to accelerate inverse design in reticular chemistry and unlock new frontiers in computational MOF discovery.
- [14] arXiv:2506.00205 [pdf, html, other]
-
Title: Unlocking the Power of Rehearsal in Continual Learning: A Theoretical PerspectiveComments: accepted to ICML 2025Subjects: Machine Learning (cs.LG)
Rehearsal-based methods have shown superior performance in addressing catastrophic forgetting in continual learning (CL) by storing and training on a subset of past data alongside new data in current task. While such a concurrent rehearsal strategy is widely used, it remains unclear if this approach is always optimal. Inspired by human learning, where sequentially revisiting tasks helps mitigate forgetting, we explore whether sequential rehearsal can offer greater benefits for CL compared to standard concurrent rehearsal. To address this question, we conduct a theoretical analysis of rehearsal-based CL in overparameterized linear models, comparing two strategies: 1) Concurrent Rehearsal, where past and new data are trained together, and 2) Sequential Rehearsal, where new data is trained first, followed by revisiting past data sequentially. By explicitly characterizing forgetting and generalization error, we show that sequential rehearsal performs better when tasks are less similar. These insights further motivate a novel Hybrid Rehearsal method, which trains similar tasks concurrently and revisits dissimilar tasks sequentially. We characterize its forgetting and generalization performance, and our experiments with deep neural networks further confirm that the hybrid approach outperforms standard concurrent rehearsal. This work provides the first comprehensive theoretical analysis of rehearsal-based CL.
- [15] arXiv:2506.00209 [pdf, html, other]
-
Title: Intercept Cancer: Cancer Pre-Screening with Large Scale Healthcare Foundation ModelsSubjects: Machine Learning (cs.LG)
Cancer screening, leading to early detection, saves lives. Unfortunately, existing screening techniques require expensive and intrusive medical procedures, not globally available, resulting in too many lost would-be-saved lives. We present CATCH-FM, CATch Cancer early with Healthcare Foundation Models, a cancer pre-screening methodology that identifies high-risk patients for further screening solely based on their historical medical records. With millions of electronic healthcare records (EHR), we establish the scaling law of EHR foundation models pretrained on medical code sequences, pretrain compute-optimal foundation models of up to 2.4 billion parameters, and finetune them on clinician-curated cancer risk prediction cohorts. In our retrospective evaluation comprising of thirty thousand patients, CATCH-FM achieved strong efficacy (60% sensitivity) with low risk (99% specificity and Negative Predictive Value), outperforming feature-based tree models as well as general and medical large language models by large margins. Despite significant demographic, healthcare system, and EHR coding differences, CATCH-FM achieves state-of-the-art pancreatic cancer risk prediction on the EHRSHOT few-shot leaderboard, outperforming EHR foundation models pretrained using on-site patient data. Our analysis demonstrates the robustness of CATCH-FM in various patient distributions, the benefits of operating in the ICD code space, and its ability to capture non-trivial cancer risk factors. Our code will be open-sourced.
- [16] arXiv:2506.00236 [pdf, html, other]
-
Title: Localized LoRA: A Structured Low-Rank Approximation for Efficient Fine-TuningSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Parameter-efficient fine-tuning (PEFT) methods, such as LoRA, offer compact and effective alternatives to full model fine-tuning by introducing low-rank updates to pretrained weights. However, most existing approaches rely on global low-rank structures, which can overlook spatial patterns spread across the parameter space. In this work, we propose Localized LoRA, a generalized framework that models weight updates as a composition of low-rank matrices applied to structured blocks of the weight matrix. This formulation enables dense, localized updates throughout the parameter space-without increasing the total number of trainable parameters. We provide a formal comparison between global, diagonal-local, and fully localized low-rank approximations, and show that our method consistently achieves lower approximation error under matched parameter budgets. Experiments on both synthetic and practical settings demonstrate that Localized LoRA offers a more expressive and adaptable alternative to existing methods, enabling efficient fine-tuning with improved performance.
- [17] arXiv:2506.00244 [pdf, html, other]
-
Title: DeGLIF for Label Noise Robust Node Classification using GNNsSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Noisy labelled datasets are generally inexpensive compared to clean labelled datasets, and the same is true for graph data. In this paper, we propose a denoising technique DeGLIF: Denoising Graph Data using Leave-One-Out Influence Function. DeGLIF uses a small set of clean data and the leave-one- out influence function to make label noise robust node-level prediction on graph data. Leave-one-out influence function approximates the change in the model parameters if a training point is removed from the training dataset. Recent advances propose a way to calculate the leave-one-out influence function for Graph Neural Networks (GNNs). We extend that recent work to estimate the change in validation loss, if a training node is removed from the training dataset. We use this estimate and a new theoretically motivated relabelling function to denoise the training dataset. We propose two DeGLIF variants to identify noisy nodes. Both these variants do not require any information about the noise model or the noise level in the dataset; DeGLIF also does not estimate these quantities. For one of these variants, we prove that the noisy points detected can indeed increase risk. We carry out detailed computational experiments on different datasets to show the effectiveness of DeGLIF. It achieves better accuracy than other baseline algorithms
- [18] arXiv:2506.00245 [pdf, html, other]
-
Title: Beyond Semantic Entropy: Boosting LLM Uncertainty Quantification with Pairwise Semantic SimilarityComments: 11 pages, 4 figures, 6 tables, link: this https URLSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Hallucination in large language models (LLMs) can be detected by assessing the uncertainty of model outputs, typically measured using entropy. Semantic entropy (SE) enhances traditional entropy estimation by quantifying uncertainty at the semantic cluster level. However, as modern LLMs generate longer one-sentence responses, SE becomes less effective because it overlooks two crucial factors: intra-cluster similarity (the spread within a cluster) and inter-cluster similarity (the distance between clusters). To address these limitations, we propose a simple black-box uncertainty quantification method inspired by nearest neighbor estimates of entropy. Our approach can also be easily extended to white-box settings by incorporating token probabilities. Additionally, we provide theoretical results showing that our method generalizes semantic entropy. Extensive empirical results demonstrate its effectiveness compared to semantic entropy across two recent LLMs (Phi3 and Llama3) and three common text generation tasks: question answering, text summarization, and machine translation. Our code is available at this https URL.
- [19] arXiv:2506.00247 [pdf, html, other]
-
Title: Performance Analysis of Convolutional Neural Network By Applying Unconstrained Binary Quadratic ProgrammingComments: 11 pages, 22 figures, accepted in IEEE COMPSAC 2025 Conference. Preprint before peer reviewSubjects: Machine Learning (cs.LG); Emerging Technologies (cs.ET)
Convolutional Neural Networks (CNNs) are pivotal in computer vision and Big Data analytics but demand significant computational resources when trained on large-scale datasets. Conventional training via back-propagation (BP) with losses like Mean Squared Error or Cross-Entropy often requires extensive iterations and may converge sub-optimally. Quantum computing offers a promising alternative by leveraging superposition, tunneling, and entanglement to search complex optimization landscapes more efficiently. In this work, we propose a hybrid optimization method that combines an Unconstrained Binary Quadratic Programming (UBQP) formulation with Stochastic Gradient Descent (SGD) to accelerate CNN training. Evaluated on the MNIST dataset, our approach achieves a 10--15\% accuracy improvement over a standard BP-CNN baseline while maintaining similar execution times. These results illustrate the potential of hybrid quantum-classical techniques in High-Performance Computing (HPC) environments for Big Data and Deep Learning. Fully realizing these benefits, however, requires a careful alignment of algorithmic structures with underlying quantum mechanisms.
- [20] arXiv:2506.00259 [pdf, html, other]
-
Title: PerFormer: A Permutation Based Vision Transformer for Remaining Useful Life PredictionSubjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
Accurately estimating the remaining useful life (RUL) for degradation systems is crucial in modern prognostic and health management (PHM). Convolutional Neural Networks (CNNs), initially developed for tasks like image and video recognition, have proven highly effectively in RUL prediction, demonstrating remarkable performance. However, with the emergence of the Vision Transformer (ViT), a Transformer model tailored for computer vision tasks such as image classification, and its demonstrated superiority over CNNs, there is a natural inclination to explore its potential in enhancing RUL prediction accuracy. Nonetheless, applying ViT directly to multivariate sensor data for RUL prediction poses challenges, primarily due to the ambiguous nature of spatial information in time series data. To address this issue, we introduce the PerFormer, a permutation-based vision transformer approach designed to permute multivariate time series data, mimicking spatial characteristics akin to image data, thereby making it suitable for ViT. To generate the desired permutation matrix, we introduce a novel permutation loss function aimed at guiding the convergence of any matrix towards a permutation matrix. Our experiments on NASA's C-MAPSS dataset demonstrate the PerFormer's superior performance in RUL prediction compared to state-of-the-art methods employing CNNs, Recurrent Neural Networks (RNNs), and various Transformer models. This underscores its effectiveness and potential in PHM applications.
- [21] arXiv:2506.00286 [pdf, html, other]
-
Title: Entropic Risk Optimization in Discounted MDPs: Sample Complexity Bounds with a Generative ModelSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Optimization and Control (math.OC); Machine Learning (stat.ML)
In this paper we analyze the sample complexities of learning the optimal state-action value function $Q^*$ and an optimal policy $\pi^*$ in a discounted Markov decision process (MDP) where the agent has recursive entropic risk-preferences with risk-parameter $\beta\neq 0$ and where a generative model of the MDP is available. We provide and analyze a simple model based approach which we call model-based risk-sensitive $Q$-value-iteration (MB-RS-QVI) which leads to $(\epsilon,\delta)$-PAC-bounds on $\|Q^*-Q^k\|$, and $\|V^*-V^{\pi_k}\|$ where $Q_k$ is the output of MB-RS-QVI after k iterations and $\pi_k$ is the greedy policy with respect to $Q_k$. Both PAC-bounds have exponential dependence on the effective horizon $\frac{1}{1-\gamma}$ and the strength of this dependence grows with the learners risk-sensitivity $|\beta|$. We also provide two lower bounds which shows that exponential dependence on $|\beta|\frac{1}{1-\gamma}$ is unavoidable in both cases. The lower bounds reveal that the PAC-bounds are both tight in $\varepsilon$ and $\delta$ and that the PAC-bound on $Q$-learning is tight in the number of actions $A$, and that the PAC-bound on policy-learning is nearly tight in $A$.
- [22] arXiv:2506.00297 [pdf, html, other]
-
Title: Improving Protein Sequence Design through Designability Preference OptimizationSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Biomolecules (q-bio.BM)
Protein sequence design methods have demonstrated strong performance in sequence generation for de novo protein design. However, as the training objective was sequence recovery, it does not guarantee designability--the likelihood that a designed sequence folds into the desired structure. To bridge this gap, we redefine the training objective by steering sequence generation toward high designability. To do this, we integrate Direct Preference Optimization (DPO), using AlphaFold pLDDT scores as the preference signal, which significantly improves the in silico design success rate. To further refine sequence generation at a finer, residue-level granularity, we introduce Residue-level Designability Preference Optimization (ResiDPO), which applies residue-level structural rewards and decouples optimization across residues. This enables direct improvement in designability while preserving regions that already perform well. Using a curated dataset with residue-level annotations, we fine-tune LigandMPNN with ResiDPO to obtain EnhancedMPNN, which achieves a nearly 3-fold increase in in silico design success rate (from 6.56% to 17.57%) on a challenging enzyme design benchmark.
- [23] arXiv:2506.00299 [pdf, html, other]
-
Title: Inference-Time Alignment of Diffusion Models with Evolutionary AlgorithmsPurvish Jajal, Nick John Eliopoulos, Benjamin Shiue-Hal Chou, George K. Thiruvathukal, James C. Davis, Yung-Hsiang LuSubjects: Machine Learning (cs.LG)
Diffusion models are state-of-the-art generative models in various domains, yet their samples often fail to satisfy downstream objectives such as safety constraints or domain-specific validity. Existing techniques for alignment require gradients, internal model access, or large computational budgets. We introduce an inference-time alignment framework based on evolutionary algorithms. We treat diffusion models as black-boxes and search their latent space to maximize alignment objectives. Our method enables efficient inference-time alignment for both differentiable and non-differentiable alignment objectives across a range of diffusion models. On the DrawBench and Open Image Preferences benchmark, our EA methods outperform state-of-the-art gradient-based and gradient-free inference-time methods. In terms of memory consumption, we require 55% to 76% lower GPU memory than gradient-based methods. In terms of running-time, we are 72% to 80% faster than gradient-based methods. We achieve higher alignment scores over 50 optimization steps on Open Image Preferences than gradient-based and gradient-free methods.
- [24] arXiv:2506.00302 [pdf, html, other]
-
Title: Beyond Atomic Geometry Representations in Materials Science: A Human-in-the-Loop Multimodal FrameworkComments: Submitted to ICML 2025 Workshop on DataWorldSubjects: Machine Learning (cs.LG); Materials Science (cond-mat.mtrl-sci)
Most materials science datasets are limited to atomic geometries (e.g., XYZ files), restricting their utility for multimodal learning and comprehensive data-centric analysis. These constraints have historically impeded the adoption of advanced machine learning techniques in the field. This work introduces MultiCrystalSpectrumSet (MCS-Set), a curated framework that expands materials datasets by integrating atomic structures with 2D projections and structured textual annotations, including lattice parameters and coordination metrics. MCS-Set enables two key tasks: (1) multimodal property and summary prediction, and (2) constrained crystal generation with partial cluster supervision. Leveraging a human-in-the-loop pipeline, MCS-Set combines domain expertise with standardized descriptors for high-quality annotation. Evaluations using state-of-the-art language and vision-language models reveal substantial modality-specific performance gaps and highlight the importance of annotation quality for generalization. MCS-Set offers a foundation for benchmarking multimodal models, advancing annotation practices, and promoting accessible, versatile materials science datasets. The dataset and implementations are available at this https URL.
- [25] arXiv:2506.00316 [pdf, html, other]
-
Title: Active Learning via Regression Beyond RealizabilitySubjects: Machine Learning (cs.LG)
We present a new active learning framework for multiclass classification based on surrogate risk minimization that operates beyond the standard realizability assumption. Existing surrogate-based active learning algorithms crucially rely on realizability$\unicode{x2014}$the assumption that the optimal surrogate predictor lies within the model class$\unicode{x2014}$limiting their applicability in practical, misspecified settings. In this work we show that under conditions significantly weaker than realizability, as long as the class of models considered is convex, one can still obtain a label and sample complexity comparable to prior work. Despite achieving similar rates, the algorithmic approaches from prior works can be shown to fail in non-realizable settings where our assumption is satisfied. Our epoch-based active learning algorithm departs from prior methods by fitting a model from the full class to the queried data in each epoch and returning an improper classifier obtained by aggregating these models.
- [26] arXiv:2506.00329 [pdf, html, other]
-
Title: Foresight: Adaptive Layer Reuse for Accelerated and High-Quality Text-to-Video GenerationSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
Diffusion Transformers (DiTs) achieve state-of-the-art results in text-to-image, text-to-video generation, and editing. However, their large model size and the quadratic cost of spatial-temporal attention over multiple denoising steps make video generation computationally expensive. Static caching mitigates this by reusing features across fixed steps but fails to adapt to generation dynamics, leading to suboptimal trade-offs between speed and quality.
We propose Foresight, an adaptive layer-reuse technique that reduces computational redundancy across denoising steps while preserving baseline performance. Foresight dynamically identifies and reuses DiT block outputs for all layers across steps, adapting to generation parameters such as resolution and denoising schedules to optimize efficiency. Applied to OpenSora, Latte, and CogVideoX, Foresight achieves up to 1.63x end-to-end speedup, while maintaining video quality. The source code of Foresight is available at \texttt{this https URL}. - [27] arXiv:2506.00337 [pdf, html, other]
-
Title: Channel-Imposed Fusion: A Simple yet Effective Method for Medical Time Series ClassificationMing Hu, Jianfu Yin, Mingyu Dou, Yuqi Wang, Ruochen Dang, Siyi Liang, Cong Hu, Yao Wang, Bingliang Hu, Quan WangSubjects: Machine Learning (cs.LG)
The automatic classification of medical time series signals, such as electroencephalogram (EEG) and electrocardiogram (ECG), plays a pivotal role in clinical decision support and early detection of diseases. Although Transformer based models have achieved notable performance by implicitly modeling temporal dependencies through self-attention mechanisms, their inherently complex architectures and opaque reasoning processes undermine their trustworthiness in high stakes clinical settings. In response to these limitations, this study shifts focus toward a modeling paradigm that emphasizes structural transparency, aligning more closely with the intrinsic characteristics of medical data. We propose a novel method, Channel Imposed Fusion (CIF), which enhances the signal-to-noise ratio through cross-channel information fusion, effectively reduces redundancy, and improves classification performance. Furthermore, we integrate CIF with the Temporal Convolutional Network (TCN), known for its structural simplicity and controllable receptive field, to construct an efficient and explicit classification framework. Experimental results on multiple publicly available EEG and ECG datasets demonstrate that the proposed method not only outperforms existing state-of-the-art (SOTA) approaches in terms of various classification metrics, but also significantly enhances the transparency of the classification process, offering a novel perspective for medical time series classification.
- [28] arXiv:2506.00356 [pdf, other]
-
Title: Exploring the Performance of Perforated Backpropagation through Further ExperimentsRorry Brenner, Evan Davis, Rushi Chaudhari, Rowan Morse, Jingyao Chen, Xirui Liu, Zhaoyi You, Laurent IttiComments: 10 pages, 7 figures, 1 tableSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Perforated Backpropagation is a neural network optimization technique based on modern understanding of the computational importance of dendrites within biological neurons. This paper explores further experiments from the original publication, generated from a hackathon held at the Carnegie Mellon Swartz Center in February 2025. Students and local Pittsburgh ML practitioners were brought together to experiment with the Perforated Backpropagation algorithm on the datasets and models which they were using for their projects. Results showed that the system could enhance their projects, with up to 90% model compression without negative impact on accuracy, or up to 16% increased accuracy of their original models.
- [29] arXiv:2506.00362 [pdf, html, other]
-
Title: FSNet: Feasibility-Seeking Neural Network for Constrained Optimization with GuaranteesSubjects: Machine Learning (cs.LG); Optimization and Control (math.OC)
Efficiently solving constrained optimization problems is crucial for numerous real-world applications, yet traditional solvers are often computationally prohibitive for real-time use. Machine learning-based approaches have emerged as a promising alternative to provide approximate solutions at faster speeds, but they struggle to strictly enforce constraints, leading to infeasible solutions in practice. To address this, we propose the Feasibility-Seeking-Integrated Neural Network (FSNet), which integrates a feasibility-seeking step directly into its solution procedure to ensure constraint satisfaction. This feasibility-seeking step solves an unconstrained optimization problem that minimizes constraint violations in a differentiable manner, enabling end-to-end training and providing guarantees on feasibility and convergence. Our experiments across a range of different optimization problems, including both smooth/nonsmooth and convex/nonconvex problems, demonstrate that FSNet can provide feasible solutions with solution quality comparable to (or in some cases better than) traditional solvers, at significantly faster speeds.
- [30] arXiv:2506.00382 [pdf, html, other]
-
Title: Spectral Insights into Data-Oblivious Critical Layers in Large Language ModelsComments: Accepted by Findings of ACL2025Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Understanding how feature representations evolve across layers in large language models (LLMs) is key to improving their interpretability and robustness. While recent studies have identified critical layers linked to specific functions or behaviors, these efforts typically rely on data-dependent analyses of fine-tuned models, limiting their use to post-hoc settings. In contrast, we introduce a data-oblivious approach to identify intrinsic critical layers in pre-fine-tuned LLMs by analyzing representation dynamics via Centered Kernel Alignment(CKA). We show that layers with significant shifts in representation space are also those most affected during fine-tuning--a pattern that holds consistently across tasks for a given model. Our spectral analysis further reveals that these shifts are driven by changes in the top principal components, which encode semantic transitions from rationales to conclusions. We further apply these findings to two practical scenarios: efficient domain adaptation, where fine-tuning critical layers leads to greater loss reduction compared to non-critical layers; and backdoor defense, where freezing them reduces attack success rates by up to 40%.
- [31] arXiv:2506.00384 [pdf, html, other]
-
Title: Deep-Learning-Driven Prefetching for Far MemorySubjects: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC); Operating Systems (cs.OS)
Modern software systems face increasing runtime performance demands, particularly in emerging architectures like far memory, where local-memory misses incur significant latency. While machine learning (ML) has proven effective in offline systems optimization, its application to high-frequency, runtime-level problems remains limited due to strict performance, generalization, and integration constraints. We present FarSight, a Linux-based far-memory system that leverages deep learning (DL) to efficiently perform accurate data prefetching. FarSight separates application semantics from runtime memory layout, allowing offline-trained DL models to predict access patterns using a compact vocabulary of ordinal possibilities, resolved at runtime through lightweight mapping structures. By combining asynchronous inference, lookahead prediction, and a cache-resident DL model, FarSight achieves high prediction accuracy with low runtime overhead. Our evaluation of FarSight on four data-intensive workloads shows that it outperforms the state-of-the-art far-memory system by up to 3.6 times. Overall, this work demonstrates the feasibility and advantages of applying modern ML techniques to complex, performance-critical software runtime problems.
- [32] arXiv:2506.00388 [pdf, html, other]
-
Title: CLARIFY: Contrastive Preference Reinforcement Learning for Untangling Ambiguous QueriesComments: ICML 2025Subjects: Machine Learning (cs.LG)
Preference-based reinforcement learning (PbRL) bypasses explicit reward engineering by inferring reward functions from human preference comparisons, enabling better alignment with human intentions. However, humans often struggle to label a clear preference between similar segments, reducing label efficiency and limiting PbRL's real-world applicability. To address this, we propose an offline PbRL method: Contrastive LeArning for ResolvIng Ambiguous Feedback (CLARIFY), which learns a trajectory embedding space that incorporates preference information, ensuring clearly distinguished segments are spaced apart, thus facilitating the selection of more unambiguous queries. Extensive experiments demonstrate that CLARIFY outperforms baselines in both non-ideal teachers and real human feedback settings. Our approach not only selects more distinguished queries but also learns meaningful trajectory embeddings.
- [33] arXiv:2506.00407 [pdf, html, other]
-
Title: Bias as a Virtue: Rethinking Generalization under Distribution ShiftsComments: 14 pagesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Machine learning models often degrade when deployed on data distributions different from their training data. Challenging conventional validation paradigms, we demonstrate that higher in-distribution (ID) bias can lead to better out-of-distribution (OOD) generalization. Our Adaptive Distribution Bridge (ADB) framework implements this insight by introducing controlled statistical diversity during training, enabling models to develop bias profiles that effectively generalize across distributions. Empirically, we observe a robust negative correlation where higher ID bias corresponds to lower OOD error--a finding that contradicts standard practices focused on minimizing validation error. Evaluation on multiple datasets shows our approach significantly improves OOD generalization. ADB achieves robust mean error reductions of up to 26.8% compared to traditional cross-validation, and consistently identifies high-performing training strategies, evidenced by percentile ranks often exceeding 74.4%. Our work provides both a practical method for improving generalization and a theoretical framework for reconsidering the role of bias in robust machine learning.
- [34] arXiv:2506.00410 [pdf, html, other]
-
Title: JojoSCL: Shrinkage Contrastive Learning for single-cell RNA sequence ClusteringSubjects: Machine Learning (cs.LG); Genomics (q-bio.GN); Machine Learning (stat.ML)
Single-cell RNA sequencing (scRNA-seq) has revolutionized our understanding of cellular processes by enabling gene expression analysis at the individual cell level. Clustering allows for the identification of cell types and the further discovery of intrinsic patterns in single-cell data. However, the high dimensionality and sparsity of scRNA-seq data continue to challenge existing clustering models. In this paper, we introduce JojoSCL, a novel self-supervised contrastive learning framework for scRNA-seq clustering. By incorporating a shrinkage estimator based on hierarchical Bayesian estimation, which adjusts gene expression estimates towards more reliable cluster centroids to reduce intra-cluster dispersion, and optimized using Stein's Unbiased Risk Estimate (SURE), JojoSCL refines both instance-level and cluster-level contrastive learning. Experiments on ten scRNA-seq datasets substantiate that JojoSCL consistently outperforms prevalent clustering methods, with further validation of its practicality through robustness analysis and ablation studies. JojoSCL's code is available at: this https URL.
- [35] arXiv:2506.00416 [pdf, html, other]
-
Title: Blockchain-Enabled Privacy-Preserving Second-Order Federated Edge Learning in Personalized HealthcareSubjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Machine Learning (stat.ML)
Federated learning (FL) has attracted increasing attention to mitigate security and privacy challenges in traditional cloud-centric machine learning models specifically in healthcare ecosystems. FL methodologies enable the training of global models through localized policies, allowing independent operations at the edge clients' level. Conventional first-order FL approaches face several challenges in personalized model training due to heterogeneous non-independent and identically distributed (non-iid) data of each edge client. Recently, second-order FL approaches maintain the stability and consistency of non-iid datasets while improving personalized model training. This study proposes and develops a verifiable and auditable optimized second-order FL framework BFEL (blockchain-enhanced federated edge learning) based on optimized FedCurv for personalized healthcare systems. FedCurv incorporates information about the importance of each parameter to each client's task (through Fisher Information Matrix) which helps to preserve client-specific knowledge and reduce model drift during aggregation. Moreover, it minimizes communication rounds required to achieve a target precision convergence for each edge client while effectively managing personalized training on non-iid and heterogeneous data. The incorporation of Ethereum-based model aggregation ensures trust, verifiability, and auditability while public key encryption enhances privacy and security. Experimental results of federated CNNs and MLPs utilizing Mnist, Cifar-10, and PathMnist demonstrate the high efficiency and scalability of the proposed framework.
- [36] arXiv:2506.00420 [pdf, other]
-
Title: A New Spatiotemporal Correlation Anomaly Detection Method that Integrates Contrastive Learning and Few-Shot Learning in Wireless Sensor NetworksSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Detecting anomalies in the data collected by WSNs can provide crucial evidence for assessing the reliability and stability of WSNs. Existing methods for WSN anomaly detection often face challenges such as the limited extraction of spatiotemporal correlation features, the absence of sample labels, few anomaly samples, and an imbalanced sample distribution. To address these issues, a spatiotemporal correlation detection model (MTAD-RD) considering both model architecture and a two-stage training strategy perspective is proposed. In terms of model structure design, the proposed MTAD-RD backbone network includes a retentive network (RetNet) enhanced by a cross-retention (CR) module, a multigranular feature fusion module, and a graph attention network module to extract internode correlation information. This proposed model can integrate the intermodal correlation features and spatial features of WSN neighbor nodes while extracting global information from time series data. Moreover, its serialized inference characteristic can remarkably reduce inference overhead. For model training, a two-stage training approach was designed. First, a contrastive learning proxy task was designed for time series data with graph structure information in WSNs, enabling the backbone network to learn transferable features from unlabeled data using unsupervised contrastive learning methods, thereby addressing the issue of missing sample labels in the dataset. Then, a caching-based sample sampler was designed to divide samples into few-shot and contrastive learning data. A specific joint loss function was developed to jointly train the dual-graph discriminator network to address the problem of sample imbalance effectively. In experiments carried out on real public datasets, the designed MTAD-RD anomaly detection method achieved an F1 score of 90.97%, outperforming existing supervised WSN anomaly detection methods.
- [37] arXiv:2506.00424 [pdf, html, other]
-
Title: COGNATE: Acceleration of Sparse Tensor Programs on Emerging Hardware using Transfer LearningChamika Sudusinghe, Gerasimos Gerogiannis Damitha Lenadora, Charles Block, Josep Torrellas, Charith MendisComments: Accepted at the 42nd International Conference on Machine LearningSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Hardware Architecture (cs.AR); Emerging Technologies (cs.ET)
Sparse tensor programs are essential in deep learning and graph analytics, driving the need for optimized processing. To meet this demand, specialized hardware accelerators are being developed. Optimizing these programs for accelerators is challenging for two reasons: program performance is highly sensitive to variations in sparse inputs, and early-stage accelerators rely on expensive simulators. Therefore, ML-based cost models used for optimizing such programs on general-purpose hardware are often ineffective for early-stage accelerators, as they require large datasets for proper training. To this end, we introduce COGNATE, a novel framework that leverages inexpensive data samples from general-purpose hardware (e.g., CPUs) to train cost models, followed by few-shot fine-tuning on emerging hardware. COGNATE exploits the homogeneity of input features across hardware platforms while effectively mitigating heterogeneity, enabling cost model training with just 5% of the data samples needed by accelerator-specific models to achieve comparable performance. We conduct extensive experiments to demonstrate that COGNATE outperforms existing techniques, achieving average speedups of 1.47x (up to 5.46x) for SpMM and 1.39x (up to 4.22x) for SDDMM.
- [38] arXiv:2506.00431 [pdf, html, other]
-
Title: TIDFormer: Exploiting Temporal and Interactive Dynamics Makes A Great Dynamic Graph TransformerComments: KDD2025Subjects: Machine Learning (cs.LG)
Due to the proficiency of self-attention mechanisms (SAMs) in capturing dependencies in sequence modeling, several existing dynamic graph neural networks (DGNNs) utilize Transformer architectures with various encoding designs to capture sequential evolutions of dynamic graphs. However, the effectiveness and efficiency of these Transformer-based DGNNs vary significantly, highlighting the importance of properly defining the SAM on dynamic graphs and comprehensively encoding temporal and interactive dynamics without extra complex modules. In this work, we propose TIDFormer, a dynamic graph TransFormer that fully exploits Temporal and Interactive Dynamics in an efficient manner. We clarify and verify the interpretability of our proposed SAM, addressing the open problem of its uninterpretable definitions on dynamic graphs in previous works. To model the temporal and interactive dynamics, respectively, we utilize the calendar-based time partitioning information and extract informative interaction embeddings for both bipartite and non-bipartite graphs using merely the sampled first-order neighbors. In addition, we jointly model temporal and interactive features by capturing potential changes in historical interaction patterns through a simple decomposition. We conduct extensive experiments on several dynamic graph datasets to verify the effectiveness and efficiency of TIDFormer. The experimental results demonstrate that TIDFormer excels, outperforming state-of-the-art models across most datasets and experimental settings. Furthermore, TIDFormer exhibits significant efficiency advantages compared to previous Transformer-based methods.
- [39] arXiv:2506.00432 [pdf, other]
-
Title: Channel Normalization for Time Series Channel IdentificationComments: ICML 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Channel identifiability (CID) refers to the ability to distinguish between individual channels in time series (TS) modeling. The absence of CID often results in producing identical outputs for identical inputs, disregarding channel-specific characteristics. In this paper, we highlight the importance of CID and propose Channel Normalization (CN), a simple yet effective normalization strategy that enhances CID by assigning distinct affine transformation parameters to each channel. We further extend CN in two ways: 1) Adaptive CN (ACN) dynamically adjusts parameters based on the input TS, improving adaptability in TS models, and 2) Prototypical CN (PCN) introduces a set of learnable prototypes instead of per-channel parameters, enabling applicability to datasets with unknown or varying number of channels and facilitating use in TS foundation models. We demonstrate the effectiveness of CN and its variants by applying them to various TS models, achieving significant performance gains for both non-CID and CID models. In addition, we analyze the success of our approach from an information theory perspective. Code is available at this https URL.
- [40] arXiv:2506.00436 [pdf, html, other]
-
Title: Learning from Double Positive and Unlabeled Data for Potential-Customer IdentificationComments: Accepted for publication in the Proceedings of IIAI AAI 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Econometrics (econ.EM); Methodology (stat.ME); Machine Learning (stat.ML)
In this study, we propose a method for identifying potential customers in targeted marketing by applying learning from positive and unlabeled data (PU learning). We consider a scenario in which a company sells a product and can observe only the customers who purchased it. Decision-makers seek to market products effectively based on whether people have loyalty to the company. Individuals with loyalty are those who are likely to remain interested in the company even without additional advertising. Consequently, those loyal customers would likely purchase from the company if they are interested in the product. In contrast, people with lower loyalty may overlook the product or buy similar products from other companies unless they receive marketing attention. Therefore, by focusing marketing efforts on individuals who are interested in the product but do not have strong loyalty, we can achieve more efficient marketing. To achieve this goal, we consider how to learn, from limited data, a classifier that identifies potential customers who (i) have interest in the product and (ii) do not have loyalty to the company. Although our algorithm comprises a single-stage optimization, its objective function implicitly contains two losses derived from standard PU learning settings. For this reason, we refer to our approach as double PU learning. We verify the validity of the proposed algorithm through numerical experiments, confirming that it functions appropriately for the problem at hand.
- [41] arXiv:2506.00437 [pdf, html, other]
-
Title: Is Your Explanation Reliable: Confidence-Aware Explanation on Graph Neural NetworksComments: In Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD25)Journal-ref: In Proceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2025)Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Explaining Graph Neural Networks (GNNs) has garnered significant attention due to the need for interpretability, enabling users to understand the behavior of these black-box models better and extract valuable insights from their predictions. While numerous post-hoc instance-level explanation methods have been proposed to interpret GNN predictions, the reliability of these explanations remains uncertain, particularly in the out-of-distribution or unknown test datasets. In this paper, we address this challenge by introducing an explainer framework with the confidence scoring module ( ConfExplainer), grounded in theoretical principle, which is generalized graph information bottleneck with confidence constraint (GIB-CC), that quantifies the reliability of generated explanations. Experimental results demonstrate the superiority of our approach, highlighting the effectiveness of the confidence score in enhancing the trustworthiness and robustness of GNN explanations.
- [42] arXiv:2506.00438 [pdf, html, other]
-
Title: PointODE: Lightweight Point Cloud Learning with Neural Ordinary Differential Equations on EdgeSubjects: Machine Learning (cs.LG); Hardware Architecture (cs.AR)
Embedded edge devices are often used as a computing platform to run real-world point cloud applications, but recent deep learning-based methods may not fit on such devices due to limited resources. In this paper, we aim to fill this gap by introducing PointODE, a parameter-efficient ResNet-like architecture for point cloud feature extraction based on a stack of MLP blocks with residual connections. We leverage Neural ODE (Ordinary Differential Equation), a continuous-depth version of ResNet originally developed for modeling the dynamics of continuous-time systems, to compress PointODE by reusing the same parameters across MLP blocks. The point-wise normalization is proposed for PointODE to handle the non-uniform distribution of feature points. We introduce PointODE-Elite as a lightweight version with 0.58M trainable parameters and design its dedicated accelerator for embedded FPGAs. The accelerator consists of a four-stage pipeline to parallelize the feature extraction for multiple points and stores the entire parameters on-chip to eliminate most of the off-chip data transfers. Compared to the ARM Cortex-A53 CPU, the accelerator implemented on a Xilinx ZCU104 board speeds up the feature extraction by 4.9x, leading to 3.7x faster inference and 3.5x better energy-efficiency. Despite the simple architecture, PointODE-Elite shows competitive accuracy to the state-of-the-art models on both synthetic and real-world classification datasets, greatly improving the trade-off between accuracy and inference cost.
- [43] arXiv:2506.00439 [pdf, html, other]
-
Title: RLAE: Reinforcement Learning-Assisted Ensemble for LLMsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Ensembling large language models (LLMs) can effectively combine diverse strengths of different models, offering a promising approach to enhance performance across various tasks. However, existing methods typically rely on fixed weighting strategies that fail to adapt to the dynamic, context-dependent characteristics of LLM capabilities. In this work, we propose Reinforcement Learning-Assisted Ensemble for LLMs (RLAE), a novel framework that reformulates LLM ensemble through the lens of a Markov Decision Process (MDP). Our approach introduces a RL agent that dynamically adjusts ensemble weights by considering both input context and intermediate generation states, with the agent being trained using rewards that directly correspond to the quality of final outputs. We implement RLAE using both single-agent and multi-agent reinforcement learning algorithms ($\text{RLAE}_\text{PPO}$ and $\text{RLAE}_\text{MAPPO}$ ), demonstrating substantial improvements over conventional ensemble methods. Extensive evaluations on a diverse set of tasks show that RLAE outperforms existing approaches by up to $3.3\%$ accuracy points, offering a more effective framework for LLM ensembling. Furthermore, our method exhibits superior generalization capabilities across different tasks without the need for retraining, while simultaneously achieving lower time latency.
- [44] arXiv:2506.00440 [pdf, html, other]
-
Title: PSI-PFL: Population Stability Index for Client Selection in non-IID Personalized Federated LearningSubjects: Machine Learning (cs.LG)
Federated Learning (FL) enables decentralized machine learning (ML) model training while preserving data privacy by keeping data localized across clients. However, non-independent and identically distributed (non-IID) data across clients poses a significant challenge, leading to skewed model updates and performance degradation. Addressing this, we propose PSI-PFL, a novel client selection framework for Personalized Federated Learning (PFL) that leverages the Population Stability Index (PSI) to quantify and mitigate data heterogeneity (so-called non-IIDness). Our approach selects more homogeneous clients based on PSI, reducing the impact of label skew, one of the most detrimental factors in FL performance. Experimental results over multiple data modalities (tabular, image, text) demonstrate that PSI-PFL significantly improves global model accuracy, outperforming state-of-the-art baselines by up to 10\% under non-IID scenarios while ensuring fairer local performance. PSI-PFL enhances FL performance and offers practical benefits in applications where data privacy and heterogeneity are critical.
- [45] arXiv:2506.00453 [pdf, html, other]
-
Title: TMetaNet: Topological Meta-Learning Framework for Dynamic Link PredictionComments: ICML2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Dynamic graphs evolve continuously, presenting challenges for traditional graph learning due to their changing structures and temporal dependencies. Recent advancements have shown potential in addressing these challenges by developing suitable meta-learning-based dynamic graph neural network models. However, most meta-learning approaches for dynamic graphs rely on fixed weight update parameters, neglecting the essential intrinsic complex high-order topological information of dynamically evolving graphs. We have designed Dowker Zigzag Persistence (DZP), an efficient and stable dynamic graph persistent homology representation method based on Dowker complex and zigzag persistence, to capture the high-order features of dynamic graphs. Armed with the DZP ideas, we propose TMetaNet, a new meta-learning parameter update model based on dynamic topological features. By utilizing the distances between high-order topological features, TMetaNet enables more effective adaptation across snapshots. Experiments on real-world datasets demonstrate TMetaNet's state-of-the-art performance and resilience to graph noise, illustrating its high potential for meta-learning and dynamic graph analysis. Our code is available at this https URL.
- [46] arXiv:2506.00457 [pdf, other]
-
Title: Revisiting LLMs as Zero-Shot Time-Series Forecasters: Small Noise Can Break Large ModelsComments: Annual Meeting of the Association for Computational Linguistics (ACL), 2025, Accepted as Short PaperSubjects: Machine Learning (cs.LG)
Large Language Models (LLMs) have shown remarkable performance across diverse tasks without domain-specific training, fueling interest in their potential for time-series forecasting. While LLMs have shown potential in zero-shot forecasting through prompting alone, recent studies suggest that LLMs lack inherent effectiveness in forecasting. Given these conflicting findings, a rigorous validation is essential for drawing reliable conclusions. In this paper, we evaluate the effectiveness of LLMs as zero-shot forecasters compared to state-of-the-art domain-specific models. Our experiments show that LLM-based zero-shot forecasters often struggle to achieve high accuracy due to their sensitivity to noise, underperforming even simple domain-specific models. We have explored solutions to reduce LLMs' sensitivity to noise in the zero-shot setting, but improving their robustness remains a significant challenge. Our findings suggest that rather than emphasizing zero-shot forecasting, a more promising direction would be to focus on fine-tuning LLMs to better process numerical sequences. Our experimental code is available at this https URL.
- [47] arXiv:2506.00458 [pdf, html, other]
-
Title: Reinforcement Learning for HanabiSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT); Multiagent Systems (cs.MA)
Hanabi has become a popular game for research when it comes to reinforcement learning (RL) as it is one of the few cooperative card games where you have incomplete knowledge of the entire environment, thus presenting a challenge for a RL agent. We explored different tabular and deep reinforcement learning algorithms to see which had the best performance both against an agent of the same type and also against other types of agents. We establish that certain agents played their highest scoring games against specific agents while others exhibited higher scores on average by adapting to the opposing agent's behavior. We attempted to quantify the conditions under which each algorithm provides the best advantage and identified the most interesting interactions between agents of different types. In the end, we found that temporal difference (TD) algorithms had better overall performance and balancing of play types compared to tabular agents. Specifically, tabular Expected SARSA and deep Q-Learning agents showed the best performance.
- [48] arXiv:2506.00459 [pdf, html, other]
-
Title: Comparing Traditional and Reinforcement-Learning Methods for Energy Storage ControlSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
We aim to better understand the tradeoffs between traditional and reinforcement learning (RL) approaches for energy storage management. More specifically, we wish to better understand the performance loss incurred when using a generative RL policy instead of using a traditional approach to find optimal control policies for specific instances. Our comparison is based on a simplified micro-grid model, that includes a load component, a photovoltaic source, and a storage device. Based on this model, we examine three use cases of increasing complexity: ideal storage with convex cost functions, lossy storage devices, and lossy storage devices with convex transmission losses. With the aim of promoting the principled use RL based methods in this challenging and important domain, we provide a detailed formulation of each use case and a detailed description of the optimization challenges. We then compare the performance of traditional and RL methods, discuss settings in which it is beneficial to use each method, and suggest avenues for future investigation.
- [49] arXiv:2506.00467 [pdf, html, other]
-
Title: SST: Self-training with Self-adaptive Thresholding for Semi-supervised LearningComments: Accepted by Information Processing & Management (IP&M)Journal-ref: Information Processing & Management, 2025, 62(5): 104158Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
Neural networks have demonstrated exceptional performance in supervised learning, benefiting from abundant high-quality annotated data. However, obtaining such data in real-world scenarios is costly and labor-intensive. Semi-supervised learning (SSL) offers a solution to this problem. Recent studies, such as Semi-ViT and Noisy Student, which employ consistency regularization or pseudo-labeling, have demonstrated significant achievements. However, they still face challenges, particularly in accurately selecting sufficient high-quality pseudo-labels due to their reliance on fixed thresholds. Recent methods such as FlexMatch and FreeMatch have introduced flexible or self-adaptive thresholding techniques, greatly advancing SSL research. Nonetheless, their process of updating thresholds at each iteration is deemed time-consuming, computationally intensive, and potentially unnecessary. To address these issues, we propose Self-training with Self-adaptive Thresholding (SST), a novel, effective, and efficient SSL framework. SST introduces an innovative Self-Adaptive Thresholding (SAT) mechanism that adaptively adjusts class-specific thresholds based on the model's learning progress. SAT ensures the selection of high-quality pseudo-labeled data, mitigating the risks of inaccurate pseudo-labels and confirmation bias. Extensive experiments demonstrate that SST achieves state-of-the-art performance with remarkable efficiency, generalization, and scalability across various architectures and datasets. Semi-SST-ViT-Huge achieves the best results on competitive ImageNet-1K SSL benchmarks, with 80.7% / 84.9% Top-1 accuracy using only 1% / 10% labeled data. Compared to the fully-supervised DeiT-III-ViT-Huge, which achieves 84.8% Top-1 accuracy using 100% labeled data, our method demonstrates superior performance using only 10% labeled data.
- [50] arXiv:2506.00476 [pdf, html, other]
-
Title: Towards Graph-Based Privacy-Preserving Federated Learning: ModelNet - A ResNet-based Model Classification DatasetComments: 8 pages, 8 figuresSubjects: Machine Learning (cs.LG)
Federated Learning (FL) has emerged as a powerful paradigm for training machine learning models across distributed data sources while preserving data locality. However, the privacy of local data is always a pivotal concern and has received a lot of attention in recent research on the FL regime. Moreover, the lack of domain heterogeneity and client-specific segregation in the benchmarks remains a critical bottleneck for rigorous evaluation. In this paper, we introduce ModelNet, a novel image classification dataset constructed from the embeddings extracted from a pre-trained ResNet50 model. First, we modify the CIFAR100 dataset into three client-specific variants, considering three domain heterogeneities (homogeneous, heterogeneous, and random). Subsequently, we train each client-specific subset of all three variants on the pre-trained ResNet50 model to save model parameters. In addition to multi-domain image data, we propose a new hypothesis to define the FL algorithm that can access the anonymized model parameters to preserve the local privacy in a more effective manner compared to existing ones. ModelNet is designed to simulate realistic FL settings by incorporating non-IID data distributions and client diversity design principles in the mainframe for both conventional and futuristic graph-driven FL algorithms. The three variants are ModelNet-S, ModelNet-D, and ModelNet-R, which are based on homogeneous, heterogeneous, and random data settings, respectively. To the best of our knowledge, we are the first to propose a cross-environment client-specific FL dataset along with the graph-based variant. Extensive experiments based on domain shifts and aggregation strategies show the effectiveness of the above variants, making it a practical benchmark for classical and graph-based FL research. The dataset and related code are available online.
- [51] arXiv:2506.00477 [pdf, html, other]
-
Title: Flashbacks to Harmonize Stability and Plasticity in Continual LearningComments: Manuscript submitted to Neural Networks (Elsevier) in August 2024; and accepted in May 2025 for publication. This version is author-accepted manuscript before copyediting and typesetting. The codes of this article will be available at this https URLSubjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (stat.ML)
We introduce Flashback Learning (FL), a novel method designed to harmonize the stability and plasticity of models in Continual Learning (CL). Unlike prior approaches that primarily focus on regularizing model updates to preserve old information while learning new concepts, FL explicitly balances this trade-off through a bidirectional form of regularization. This approach effectively guides the model to swiftly incorporate new knowledge while actively retaining its old knowledge. FL operates through a two-phase training process and can be seamlessly integrated into various CL methods, including replay, parameter regularization, distillation, and dynamic architecture techniques. In designing FL, we use two distinct knowledge bases: one to enhance plasticity and another to improve stability. FL ensures a more balanced model by utilizing both knowledge bases to regularize model updates. Theoretically, we analyze how the FL mechanism enhances the stability-plasticity balance. Empirically, FL demonstrates tangible improvements over baseline methods within the same training budget. By integrating FL into at least one representative baseline from each CL category, we observed an average accuracy improvement of up to 4.91% in Class-Incremental and 3.51% in Task-Incremental settings on standard image classification benchmarks. Additionally, measurements of the stability-to-plasticity ratio confirm that FL effectively enhances this balance. FL also outperforms state-of-the-art CL methods on more challenging datasets like ImageNet.
- [52] arXiv:2506.00478 [pdf, html, other]
-
Title: Dynamic Domain Adaptation-Driven Physics-Informed Graph Representation Learning for AC-OPFHongjie Zhu, Zezheng Zhang, Zeyu Zhang, Yu Bai, Shimin Wen, Huazhang Wang, Daji Ergu, Ying Cai, Yang ZhaoSubjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
Alternating Current Optimal Power Flow (AC-OPF) aims to optimize generator power outputs by utilizing the non-linear relationships between voltage magnitudes and phase angles in a power system. However, current AC-OPF solvers struggle to effectively represent the complex relationship between variable distributions in the constraint space and their corresponding optimal solutions. This limitation in constraint modeling restricts the system's ability to develop diverse knowledge representations. Additionally, modeling the power grid solely based on spatial topology further limits the integration of additional prior knowledge, such as temporal information. To overcome these challenges, we propose DDA-PIGCN (Dynamic Domain Adaptation-Driven Physics-Informed Graph Convolutional Network), a new method designed to address constraint-related issues and build a graph-based learning framework that incorporates spatiotemporal features. DDA-PIGCN improves consistency optimization for features with varying long-range dependencies by applying multi-layer, hard physics-informed constraints. It also uses a dynamic domain adaptation learning mechanism that iteratively updates and refines key state variables under predefined constraints, enabling precise constraint verification. Moreover, it captures spatiotemporal dependencies between generators and loads by leveraging the physical structure of the power grid, allowing for deep integration of topological information across time and space. Extensive comparative and ablation studies show that DDA-PIGCN delivers strong performance across several IEEE standard test cases (such as case9, case30, and case300), achieving mean absolute errors (MAE) from 0.0011 to 0.0624 and constraint satisfaction rates between 99.6% and 100%, establishing it as a reliable and efficient AC-OPF solver.
- [53] arXiv:2506.00482 [pdf, other]
-
Title: BenchHub: A Unified Benchmark Suite for Holistic and Customizable LLM EvaluationSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
As large language models (LLMs) continue to advance, the need for up-to-date and well-organized benchmarks becomes increasingly critical. However, many existing datasets are scattered, difficult to manage, and make it challenging to perform evaluations tailored to specific needs or domains, despite the growing importance of domain-specific models in areas such as math or code. In this paper, we introduce BenchHub, a dynamic benchmark repository that empowers researchers and developers to evaluate LLMs more effectively. BenchHub aggregates and automatically classifies benchmark datasets from diverse domains, integrating 303K questions across 38 benchmarks. It is designed to support continuous updates and scalable data management, enabling flexible and customizable evaluation tailored to various domains or use cases. Through extensive experiments with various LLM families, we demonstrate that model performance varies significantly across domain-specific subsets, emphasizing the importance of domain-aware benchmarking. We believe BenchHub can encourage better dataset reuse, more transparent model comparisons, and easier identification of underrepresented areas in existing benchmarks, offering a critical infrastructure for advancing LLM evaluation research.
- [54] arXiv:2506.00486 [pdf, html, other]
-
Title: It Takes a Good Model to Train a Good Model: Generalized Gaussian Priors for Optimized LLMsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Despite rapid advancements in the research and deployment of large language models (LLMs), the statistical distribution of model parameters, as well as their influence on initialization, training dynamics, and downstream efficiency, has received surprisingly little attention. A recent work introduced BackSlash, a training-time compression algorithm. It first demonstrated that pre-trained LLM parameters follow generalized Gaussian distributions (GGDs) better. By optimizing GG priors during training, BackSlash can reduce parameters by up to 90\% with minimal performance loss. Building on this foundational insight, we propose a unified, end-to-end framework for LLM optimization based on the GG model. Our contributions are threefold: (1) GG-based initialization scheme that aligns with the statistical structure of trained models, resulting in faster convergence and improved accuracy; (2) DeepShape, a post-training regularization method that reshapes weight distributions to match a GG profile, improving compressibility with minimized degradation in performance; and (3) RF8, a compact and hardware-efficient 8-bit floating-point format designed for GG-distributed-initialized BackSlash training, enabling low-cost inference without compromising accuracy. Experiments across diverse model architectures show that our framework consistently yields smaller and faster models that match or outperform standard training baselines. By grounding LLM development in principled statistical modeling, this work forges a new path toward efficient, scalable, and hardware-aware AI systems. The code is available on our project page: this https URL.
- [55] arXiv:2506.00495 [pdf, html, other]
-
Title: FLoE: Fisher-Based Layer Selection for Efficient Sparse Adaptation of Low-Rank ExpertsComments: 17 pages, 9 figuresSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL); Machine Learning (stat.ML)
Parameter-Efficient Fine-Tuning (PEFT) methods have emerged as a widely adopted strategy for adapting pre-trained Large Language Models (LLMs) to downstream tasks, significantly reducing memory and computational costs. However, most existing PEFT techniques uniformly deploy LoRA adapters across all layers, disregarding the intrinsic heterogeneity of layer contributions and task-specific rank requirements. This uniform paradigm leads to redundant parameter allocation and suboptimal adaptation efficiency. To address these limitations, we propose FLoE, a novel PEFT framework that introduces two key innovations: (i) a Fisher information-guided importance scoring mechanism to dynamically identify task-critical transformer layers for MoE-based low-rank adaptation, enabling sparse adapter deployment; and (ii) a Bayesian optimization-driven rank allocator that automatically determines optimal LoRA ranks on specific datasets without exhaustive grid search. Extensive experiments across diverse LLMs and benchmarks reveal that FLoE achieves impressive efficiency-accuracy trade-offs, making FLoE particularly advantageous in resource-constrained environments that necessitate rapid adaptation.
- [56] arXiv:2506.00499 [pdf, html, other]
-
Title: Federated learning framework for collaborative remaining useful life prognostics: an aircraft engine case studySubjects: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC); Emerging Technologies (cs.ET); Systems and Control (eess.SY); Machine Learning (stat.ML)
Complex systems such as aircraft engines are continuously monitored by sensors. In predictive aircraft maintenance, the collected sensor measurements are used to estimate the health condition and the Remaining Useful Life (RUL) of such systems. However, a major challenge when developing prognostics is the limited number of run-to-failure data samples. This challenge could be overcome if multiple airlines would share their run-to-failure data samples such that sufficient learning can be achieved. Due to privacy concerns, however, airlines are reluctant to share their data in a centralized setting. In this paper, a collaborative federated learning framework is therefore developed instead. Here, several airlines cooperate to train a collective RUL prognostic machine learning model, without the need to centrally share their data. For this, a decentralized validation procedure is proposed to validate the prognostics model without sharing any data. Moreover, sensor data is often noisy and of low quality. This paper therefore proposes four novel methods to aggregate the parameters of the global prognostic model. These methods enhance the robustness of the FL framework against noisy data. The proposed framework is illustrated for training a collaborative RUL prognostic model for aircraft engines, using the N-CMAPSS dataset. Here, six airlines are considered, that collaborate in the FL framework to train a collective RUL prognostic model for their aircraft's engines. When comparing the proposed FL framework with the case where each airline independently develops their own prognostic model, the results show that FL leads to more accurate RUL prognostics for five out of the six airlines. Moreover, the novel robust aggregation methods render the FL framework robust to noisy data samples.
- [57] arXiv:2506.00505 [pdf, html, other]
-
Title: From Rules to Rewards: Reinforcement Learning for Interest Rate Adjustment in DeFi LendingSubjects: Machine Learning (cs.LG)
Decentralized Finance (DeFi) lending enables permissionless borrowing via smart contracts. However, it faces challenges in optimizing interest rates, mitigating bad debt, and improving capital efficiency. Rule-based interest-rate models struggle to adapt to dynamic market conditions, leading to inefficiencies. This work applies Offline Reinforcement Learning (RL) to optimize interest rate adjustments in DeFi lending protocols. Using historical data from Aave protocol, we evaluate three RL approaches: Conservative Q-Learning (CQL), Behavior Cloning (BC), and TD3 with Behavior Cloning (TD3-BC). TD3-BC demonstrates superior performance in balancing utilization, capital stability, and risk, outperforming existing models. It adapts effectively to historical stress events like the May 2021 crash and the March 2023 USDC depeg, showcasing potential for automated, real-time governance.
- [58] arXiv:2506.00528 [pdf, html, other]
-
Title: Ultra-Quantisation: Efficient Embedding Search via 1.58-bit EncodingsComments: Submitted to SISAP25 International Conference on Similarity Search and ApplicationsSubjects: Machine Learning (cs.LG); Databases (cs.DB)
Many modern search domains comprise high-dimensional vectors of floating point numbers derived from neural networks, in the form of embeddings. Typical embeddings range in size from hundreds to thousands of dimensions, making the size of the embeddings, and the speed of comparison, a significant issue.
Quantisation is a class of mechanism which replaces the floating point values with a smaller representation, for example a short integer. This gives an approximation of the embedding space in return for a smaller data representation and a faster comparison function.
Here we take this idea almost to its extreme: we show how vectors of arbitrary-precision floating point values can be replaced by vectors whose elements are drawn from the set {-1,0,1}. This yields very significant savings in space and metric evaluation cost, while maintaining a strong correlation for similarity measurements.
This is achieved by way of a class of convex polytopes which exist in the high-dimensional space. In this article we give an outline description of these objects, and show how they can be used for the basis of such radical quantisation while maintaining a surprising degree of accuracy. - [59] arXiv:2506.00531 [pdf, html, other]
-
Title: M2WLLM: Multi-Modal Multi-Task Ultra-Short-term Wind Power Prediction Algorithm Based on Large Language ModelSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
The integration of wind energy into power grids necessitates accurate ultra-short-term wind power forecasting to ensure grid stability and optimize resource allocation. This study introduces M2WLLM, an innovative model that leverages the capabilities of Large Language Models (LLMs) for predicting wind power output at granular time intervals. M2WLLM overcomes the limitations of traditional and deep learning methods by seamlessly integrating textual information and temporal numerical data, significantly improving wind power forecasting accuracy through multi-modal data. Its architecture features a Prompt Embedder and a Data Embedder, enabling an effective fusion of textual prompts and numerical inputs within the LLMs framework. The Semantic Augmenter within the Data Embedder translates temporal data into a format that the LLMs can comprehend, enabling it to extract latent features and improve prediction accuracy. The empirical evaluations conducted on wind farm data from three Chinese provinces demonstrate that M2WLLM consistently outperforms existing methods, such as GPT4TS, across various datasets and prediction horizons. The results highlight LLMs' ability to enhance accuracy and robustness in ultra-short-term forecasting and showcase their strong few-shot learning capabilities.
- [60] arXiv:2506.00533 [pdf, html, other]
-
Title: RsGCN: Rescaling Enhances Generalization of GCNs for Solving Scalable Traveling Salesman ProblemsSubjects: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
Neural traveling salesman problem (TSP) solvers face two critical challenges: poor generalization for scalable TSPs and high training costs. To address these challenges, we propose a new Rescaling Graph Convolutional Network (RsGCN). Focusing on the scale-dependent features (i.e., features varied with problem scales) related to nodes and edges that influence the sensitivity of GCNs to the problem scales, a Rescaling Mechanism in RsGCN enhances the generalization capability by (1) rescaling adjacent nodes to construct a subgraph with a uniform number of adjacent nodes for each node across various scales of TSPs, which stabilizes the graph message aggregation; (2) rescaling subgraph edges to adjust the lengths of subgraph edges to the same magnitude, which maintains numerical consistency. In addition, an efficient training strategy with a mixed-scale dataset and bidirectional loss is used in RsGCN. To fully exploit the heatmaps generated by RsGCN, we design an efficient post-search algorithm termed Re2Opt, in which a reconstruction process based on adaptive weight is incorporated to help avoid local optima. Based on a combined architecture of RsGCN and Re2Opt, our solver achieves remarkable generalization and low training cost: with only 3 epochs of training on the mixed-scale dataset containing instances with up to 100 nodes, it can be generalized successfully to 10K-node instances without any fine-tuning. Extensive experiments demonstrate our state-of-the-art performance across uniform distribution instances of 9 different scales from 20 to 10K nodes and 78 real-world instances from TSPLIB, while requiring the fewest learnable parameters and training epochs among neural competitors.
- [61] arXiv:2506.00545 [pdf, html, other]
-
Title: Imputation of Missing Data in Smooth Pursuit Eye Movements Using a Self-Attention-based Deep Learning ApproachComments: 17 pages, 10 figures, 3 tablesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Missing data is a relevant issue in time series, especially in biomedical sequences such as those corresponding to smooth pursuit eye movements, which often contain gaps due to eye blinks and track losses, complicating the analysis and extraction of meaningful biomarkers. In this paper, a novel imputation framework is proposed using Self-Attention-based Imputation networks for time series, which leverages the power of deep learning and self-attention mechanisms to impute missing data. We further refine the imputed data using a custom made autoencoder, tailored to represent smooth pursuit eye movement sequences. The proposed approach was implemented using 5,504 sequences from 172 Parkinsonian patients and healthy controls. Results show a significant improvement in the accuracy of reconstructed eye movement sequences with respect to other state of the art techniques, substantially reducing the values for common time domain error metrics such as the mean absolute error, mean relative error, and root mean square error, while also preserving the signal's frequency domain characteristics. Moreover, it demonstrates robustness when large intervals of data are missing. This method offers an alternative solution for robustly handling missing data in time series, enhancing the reliability of smooth pursuit analysis for the screening and monitoring of neurodegenerative disorders.
- [62] arXiv:2506.00555 [pdf, html, other]
-
Title: MMedAgent-RL: Optimizing Multi-Agent Collaboration for Multimodal Medical ReasoningPeng Xia, Jinglu Wang, Yibo Peng, Kaide Zeng, Xian Wu, Xiangru Tang, Hongtu Zhu, Yun Li, Shujie Liu, Yan Lu, Huaxiu YaoSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
Medical Large Vision-Language Models (Med-LVLMs) have shown strong potential in multimodal diagnostic tasks. However, existing single-agent models struggle to generalize across diverse medical specialties, limiting their performance. Recent efforts introduce multi-agent collaboration frameworks inspired by clinical workflows, where general practitioners (GPs) and specialists interact in a fixed sequence. Despite improvements, these static pipelines lack flexibility and adaptability in reasoning. To address this, we propose MMedAgent-RL, a reinforcement learning (RL)-based multi-agent framework that enables dynamic, optimized collaboration among medical agents. Specifically, we train two GP agents based on Qwen2.5-VL via RL: the triage doctor learns to assign patients to appropriate specialties, while the attending physician integrates the judgments from multi-specialists and its own knowledge to make final decisions. To address the inconsistency in specialist outputs, we introduce a curriculum learning (CL)-guided RL strategy that progressively teaches the attending physician to balance between imitating specialists and correcting their mistakes. Experiments on five medical VQA benchmarks demonstrate that MMedAgent-RL not only outperforms both open-source and proprietary Med-LVLMs, but also exhibits human-like reasoning patterns. Notably, it achieves an average performance gain of 18.4% over supervised fine-tuning baselines.
- [63] arXiv:2506.00563 [pdf, html, other]
-
Title: Understanding Behavioral Metric Learning: A Large-Scale Study on Distracting Reinforcement Learning EnvironmentsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
A key approach to state abstraction is approximating behavioral metrics (notably, bisimulation metrics) in the observation space and embedding these learned distances in the representation space. While promising for robustness to task-irrelevant noise, as shown in prior work, accurately estimating these metrics remains challenging, requiring various design choices that create gaps between theory and practice. Prior evaluations focus mainly on final returns, leaving the quality of learned metrics and the source of performance gains unclear. To systematically assess how metric learning works in deep reinforcement learning (RL), we evaluate five recent approaches, unified conceptually as isometric embeddings with varying design choices. We benchmark them with baselines across 20 state-based and 14 pixel-based tasks, spanning 370 task configurations with diverse noise settings. Beyond final returns, we introduce the evaluation of a denoising factor to quantify the encoder's ability to filter distractions. To further isolate the effect of metric learning, we propose and evaluate an isolated metric estimation setting, in which the encoder is influenced solely by the metric loss. Finally, we release an open-source, modular codebase to improve reproducibility and support future research on metric learning in deep RL.
- [64] arXiv:2506.00569 [pdf, html, other]
-
Title: AutoMixAlign: Adaptive Data Mixing for Multi-Task Preference Optimization in LLMsNicholas E. Corrado, Julian Katz-Samuels, Adithya Devraj, Hyokun Yun, Chao Zhang, Yi Xu, Yi Pan, Bing Yin, Trishul ChilimbiComments: ACL 2025, Main ConferenceSubjects: Machine Learning (cs.LG)
When aligning large language models (LLMs), their performance on various tasks (such as being helpful, harmless, and honest) depends heavily on the composition of their training data. However, selecting a data mixture that achieves strong performance across all tasks is challenging. Existing approaches rely on large ablation studies, heuristics, or human intuition, but these can be prohibitively expensive and suboptimal. We study this problem in the setting of preference optimization via DPO and introduce AutoMixAlign (AMA), a theoretically-grounded algorithm that adaptively mixes datasets during training to balance performance across tasks. AMA first trains \textit{specialist models} for each task to determine losses that correspond to strong task performance. Then, it trains a generalist model using a novel minimax optimization that prioritizes tasks for which generalist model losses deviate most from specialist model losses. To optimize this problem, we propose two algorithms: (1) AMA-R, which adaptively reweights the objective to prioritize tasks, and (2) AMA-S, which adaptively adjusts how much data is sampled from each task to prioritize tasks. Both algorithms achieve a convergence rate of $O(1/\sqrt{T})$ in the convex case. AMA-R's convergence result follows from Sagawa et al. (2019), and we provide a convergence proof for AMA-S using online learning techniques such as EXP3. We evaluate AMA on several multitask alignment setups and find that AMA outperforms the standard alignment approach -- which simply optimizes the total loss across all tasks -- and also outperforms model merging methods.
- [65] arXiv:2506.00573 [pdf, html, other]
-
Title: Neural Estimation for Scaling Entropic Multimarginal Optimal TransportSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Multimarginal optimal transport (MOT) is a powerful framework for modeling interactions between multiple distributions, yet its applicability is bottlenecked by a high computational overhead. Entropic regularization provides computational speedups via the multimarginal Sinkhorn algorithm, whose time complexity, for a dataset size $n$ and $k$ marginals, generally scales as $O(n^k)$. However, this dependence on the dataset size $n$ is computationally prohibitive for many machine learning problems. In this work, we propose a new computational framework for entropic MOT, dubbed Neural Entropic MOT (NEMOT), that enjoys significantly improved scalability. NEMOT employs neural networks trained using mini-batches, which transfers the computational complexity from the dataset size to the size of the mini-batch, leading to substantial gains. We provide formal guarantees on the accuracy of NEMOT via non-asymptotic error bounds. We supplement these with numerical results that demonstrate the performance gains of NEMOT over Sinkhorn's algorithm, as well as extensions to neural computation of multimarginal entropic Gromov-Wasserstein alignment. In particular, orders-of-magnitude speedups are observed relative to the state-of-the-art, with a notable increase in the feasible number of samples and marginals. NEMOT seamlessly integrates as a module in large-scale machine learning pipelines, and can serve to expand the practical applicability of entropic MOT for tasks involving multimarginal data.
- [66] arXiv:2506.00574 [pdf, html, other]
-
Title: Prompt-Tuned LLM-Augmented DRL for Dynamic O-RAN Network SlicingSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Modern wireless networks must adapt to dynamic conditions while efficiently managing diverse service demands. Traditional deep reinforcement learning (DRL) struggles in these environments, as scattered and evolving feedback makes optimal decision-making challenging. Large Language Models (LLMs) offer a solution by structuring unorganized network feedback into meaningful latent representations, helping RL agents recognize patterns more effectively. For example, in O-RAN slicing, concepts like SNR, power levels and throughput are semantically related, and LLMs can naturally cluster them, providing a more interpretable state representation. To leverage this capability, we introduce a contextualization-based adaptation method that integrates learnable prompts into an LLM-augmented DRL framework. Instead of relying on full model fine-tuning, we refine state representations through task-specific prompts that dynamically adjust to network conditions. Utilizing ORANSight, an LLM trained on O-RAN knowledge, we develop Prompt-Augmented Multi agent RL (PA-MRL) framework. Learnable prompts optimize both semantic clustering and RL objectives, allowing RL agents to achieve higher rewards in fewer iterations and adapt more efficiently. By incorporating prompt-augmented learning, our approach enables faster, more scalable, and adaptive resource allocation in O-RAN slicing. Experimental results show that it accelerates convergence and outperforms other baselines.
- [67] arXiv:2506.00576 [pdf, html, other]
-
Title: ORAN-GUIDE: RAG-Driven Prompt Learning for LLM-Augmented Reinforcement Learning in O-RAN Network SlicingSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Advanced wireless networks must support highly dynamic and heterogeneous service demands. Open Radio Access Network (O-RAN) architecture enables this flexibility by adopting modular, disaggregated components, such as the RAN Intelligent Controller (RIC), Centralized Unit (CU), and Distributed Unit (DU), that can support intelligent control via machine learning (ML). While deep reinforcement learning (DRL) is a powerful tool for managing dynamic resource allocation and slicing, it often struggles to process raw, unstructured input like RF features, QoS metrics, and traffic trends. These limitations hinder policy generalization and decision efficiency in partially observable and evolving environments. To address this, we propose \textit{ORAN-GUIDE}, a dual-LLM framework that enhances multi-agent RL (MARL) with task-relevant, semantically enriched state representations. The architecture employs a domain-specific language model, ORANSight, pretrained on O-RAN control and configuration data, to generate structured, context-aware prompts. These prompts are fused with learnable tokens and passed to a frozen GPT-based encoder that outputs high-level semantic representations for DRL agents. This design adopts a retrieval-augmented generation (RAG) style pipeline tailored for technical decision-making in wireless systems. Experimental results show that ORAN-GUIDE improves sample efficiency, policy convergence, and performance generalization over standard MARL and single-LLM baselines.
- [68] arXiv:2506.00580 [pdf, html, other]
-
Title: Slow Feature Analysis as Variational Inference ObjectiveSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
This work presents a novel probabilistic interpretation of Slow Feature Analysis (SFA) through the lens of variational inference. Unlike prior formulations that recover linear SFA from Gaussian state-space models with linear emissions, this approach relaxes the key constraint of linearity. While it does not lead to full equivalence to non-linear SFA, it recasts the classical slowness objective in a variational framework. Specifically, it allows the slowness objective to be interpreted as a regularizer to a reconstruction loss. Furthermore, we provide arguments, why -- from the perspective of slowness optimization -- the reconstruction loss takes on the role of the constraints that ensure informativeness in SFA. We conclude with a discussion of potential new research directions.
- [69] arXiv:2506.00587 [pdf, html, other]
-
Title: Decoding the Stressed Brain with Geometric Machine LearningComments: 12 pages, 5 figures. This version has been accepted as a full paper at the 2025 AI in Healthcare (AIiH) ConferenceSubjects: Machine Learning (cs.LG)
Stress significantly contributes to both mental and physical disorders, yet traditional self-reported questionnaires are inherently subjective. In this study, we introduce a novel framework that employs geometric machine learning to detect stress from raw EEG recordings. Our approach constructs graphs by integrating structural connectivity (derived from electrode spatial arrangement) with functional connectivity from pairwise signal correlations. A spatio-temporal graph convolutional network (ST-GCN) processes these graphs to capture spatial and temporal dynamics. Experiments on the SAM-40 dataset show that the ST-GCN outperforms standard machine learning models on all key classification metrics and enhances interpretability, explored through ablation analyses of key channels and brain regions. These results pave the way for more objective and accurate stress detection methods.
- [70] arXiv:2506.00588 [pdf, html, other]
-
Title: Temporal Chunking Enhances Recognition of Implicit Sequential PatternsJayanta Dey, Nicholas Soures, Miranda Gonzales, Itamar Lerner, Christopher Kanan, Dhireesha KudithipudiSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
In this pilot study, we propose a neuro-inspired approach that compresses temporal sequences into context-tagged chunks, where each tag represents a recurring structural unit or``community'' in the sequence. These tags are generated during an offline sleep phase and serve as compact references to past experience, allowing the learner to incorporate information beyond its immediate input range. We evaluate this idea in a controlled synthetic environment designed to reveal the limitations of traditional neural network based sequence learners, such as recurrent neural networks (RNNs), when facing temporal patterns on multiple timescales. We evaluate this idea in a controlled synthetic environment designed to reveal the limitations of traditional neural network based sequence learners, such as recurrent neural networks (RNNs), when facing temporal patterns on multiple timescales. Our results, while preliminary, suggest that temporal chunking can significantly enhance learning efficiency under resource constrained settings. A small-scale human pilot study using a Serial Reaction Time Task further motivates the idea of structural abstraction. Although limited to synthetic tasks, this work serves as an early proof-of-concept, with initial evidence that learned context tags can transfer across related task, offering potential for future applications in transfer learning.
- [71] arXiv:2506.00592 [pdf, html, other]
-
Title: Mitigating Plasticity Loss in Continual Reinforcement Learning by Reducing ChurnComments: Accepted to ICML 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Plasticity, or the ability of an agent to adapt to new tasks, environments, or distributions, is crucial for continual learning. In this paper, we study the loss of plasticity in deep continual RL from the lens of churn: network output variability for out-of-batch data induced by mini-batch training. We demonstrate that (1) the loss of plasticity is accompanied by the exacerbation of churn due to the gradual rank decrease of the Neural Tangent Kernel (NTK) matrix; (2) reducing churn helps prevent rank collapse and adjusts the step size of regular RL gradients adaptively. Moreover, we introduce Continual Churn Approximated Reduction (C-CHAIN) and demonstrate it improves learning performance and outperforms baselines in a diverse range of continual learning environments on OpenAI Gym Control, ProcGen, DeepMind Control Suite, and MinAtar benchmarks.
- [72] arXiv:2506.00594 [pdf, html, other]
-
Title: Graph Evidential Learning for Anomaly DetectionComments: Accepted by KDD25Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Graph anomaly detection faces significant challenges due to the scarcity of reliable anomaly-labeled datasets, driving the development of unsupervised methods. Graph autoencoders (GAEs) have emerged as a dominant approach by reconstructing graph structures and node features while deriving anomaly scores from reconstruction errors. However, relying solely on reconstruction error for anomaly detection has limitations, as it increases the sensitivity to noise and overfitting. To address these issues, we propose Graph Evidential Learning (GEL), a probabilistic framework that redefines the reconstruction process through evidential learning. By modeling node features and graph topology using evidential distributions, GEL quantifies two types of uncertainty: graph uncertainty and reconstruction uncertainty, incorporating them into the anomaly scoring mechanism. Extensive experiments demonstrate that GEL achieves state-of-the-art performance while maintaining high robustness against noise and structural perturbations.
- [73] arXiv:2506.00614 [pdf, html, other]
-
Title: Predictability-Aware Compression and Decompression Framework for Multichannel Time Series DataComments: 17 pages,3 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Real-world multichannel time series prediction faces growing demands for efficiency across edge and cloud environments, making channel compression a timely and essential problem. Motivated by success of Multiple-Input Multiple-Output (MIMO) methods, we propose a predictability-aware compression-decompression framework to reduce runtime, lower communication cost, and maintain prediction accuracy across diverse predictors. The core idea involves using a circular periodicity key matrix with orthogonality to capture underlying time series predictability during compression and to mitigate reconstruction errors during decompression by relaxing oversimplified data assumptions. Theoretical and empirical analyses show that the proposed framework is both time-efficient and scalable under a large number of channels. Extensive experiments on six datasets across various predictors demonstrate that the proposed method achieves superior overall performance by jointly considering prediction accuracy and runtime, while maintaining strong compatibility with diverse predictors.
- [74] arXiv:2506.00620 [pdf, html, other]
-
Title: Model Reprogramming Demystified: A Neural Tangent Kernel PerspectiveMing-Yu Chung, Jiashuo Fan, Hancheng Ye, Qinsi Wang, Wei-Chen Shen, Chia-Mu Yu, Pin-Yu Chen, Sy-Yen KuoComments: 24 pages, 8 figures, 4 tablesSubjects: Machine Learning (cs.LG)
Model Reprogramming (MR) is a resource-efficient framework that adapts large pre-trained models to new tasks with minimal additional parameters and data, offering a promising solution to the challenges of training large models for diverse tasks. Despite its empirical success across various domains such as computer vision and time-series forecasting, the theoretical foundations of MR remain underexplored. In this paper, we present a comprehensive theoretical analysis of MR through the lens of the Neural Tangent Kernel (NTK) framework. We demonstrate that the success of MR is governed by the eigenvalue spectrum of the NTK matrix on the target dataset and establish the critical role of the source model's effectiveness in determining reprogramming outcomes. Our contributions include a novel theoretical framework for MR, insights into the relationship between source and target models, and extensive experiments validating our findings.
- [75] arXiv:2506.00630 [pdf, html, other]
-
Title: Probabilistic Forecasting for Building Energy Systems using Time-Series Foundation ModelsYoung Jin Park, Francois Germain, Jing Liu, Ye Wang, Toshiaki Koike-Akino, Gordon Wichern, Navid Azizan, Christopher R. Laughman, Ankush ChakrabartyComments: Preliminary version appeared in NeurIPS TSALM Workshop: this https URLSubjects: Machine Learning (cs.LG)
Decision-making in building energy systems critically depends on the predictive accuracy of relevant time-series models. In scenarios lacking extensive data from a target building, foundation models (FMs) represent a promising technology that can leverage prior knowledge from vast and diverse pre-training datasets to construct accurate probabilistic predictors for use in decision-making tools. This paper investigates the applicability and fine-tuning strategies of time-series foundation models (TSFMs) in building energy forecasting. We analyze both full fine-tuning and parameter-efficient fine-tuning approaches, particularly low-rank adaptation (LoRA), by using real-world data from a commercial net-zero energy building to capture signals such as room occupancy, carbon emissions, plug loads, and HVAC energy consumption. Our analysis reveals that the zero-shot predictive performance of TSFMs is generally suboptimal. To address this shortcoming, we demonstrate that employing either full fine-tuning or parameter-efficient fine-tuning significantly enhances forecasting accuracy, even with limited historical data. Notably, fine-tuning with low-rank adaptation (LoRA) substantially reduces computational costs without sacrificing accuracy. Furthermore, fine-tuned TSFMs consistently outperform state-of-the-art deep forecasting models (e.g., temporal fusion transformers) in accuracy, robustness, and generalization across varying building zones and seasonal conditions. These results underline the efficacy of TSFMs for practical, data-constrained building energy management systems, enabling improved decision-making in pursuit of energy efficiency and sustainability.
- [76] arXiv:2506.00635 [pdf, html, other]
-
Title: Learning with Calibration: Exploring Test-Time Computing of Spatio-Temporal ForecastingComments: 28 pages, 9 figures, 8 tablesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Machine Learning (stat.ML)
Spatio-temporal forecasting is crucial in many domains, such as transportation, meteorology, and energy. However, real-world scenarios frequently present challenges such as signal anomalies, noise, and distributional shifts. Existing solutions primarily enhance robustness by modifying network architectures or training procedures. Nevertheless, these approaches are computationally intensive and resource-demanding, especially for large-scale applications. In this paper, we explore a novel test-time computing paradigm, namely learning with calibration, ST-TTC, for spatio-temporal forecasting. Through learning with calibration, we aim to capture periodic structural biases arising from non-stationarity during the testing phase and perform real-time bias correction on predictions to improve accuracy. Specifically, we first introduce a spectral-domain calibrator with phase-amplitude modulation to mitigate periodic shift and then propose a flash updating mechanism with a streaming memory queue for efficient test-time computation. ST-TTC effectively bypasses complex training-stage techniques, offering an efficient and generalizable paradigm. Extensive experiments on real-world datasets demonstrate the effectiveness, universality, flexibility and efficiency of our proposed method.
- [77] arXiv:2506.00642 [pdf, html, other]
-
Title: Rethinking Neural-based Matrix Inversion: Why can't, and Where canSubjects: Machine Learning (cs.LG)
Deep neural networks have achieved substantial success across various scientific computing tasks. A pivotal challenge within this domain is the rapid and parallel approximation of matrix inverses, critical for numerous applications. Despite significant progress, there currently exists no universal neural-based method for approximating matrix inversion. This paper presents a theoretical analysis demonstrating the fundamental limitations of neural networks in developing a general matrix inversion model. We expand the class of Lipschitz functions to encompass a wider array of neural network models, thereby refining our theoretical approach. Moreover, we delineate specific conditions under which neural networks can effectively approximate matrix inverses. Our theoretical results are supported by experimental results from diverse matrix datasets, exploring the efficacy of neural networks in addressing the matrix inversion challenge.
- [78] arXiv:2506.00653 [pdf, html, other]
-
Title: Linear Representation Transferability Hypothesis: Leveraging Small Models to Steer Large ModelsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
It has been hypothesized that neural networks with similar architectures trained on similar data learn shared representations relevant to the learning task. We build on this idea by extending the conceptual framework where representations learned across models trained on the same data can be expressed as linear combinations of a \emph{universal} set of basis features. These basis features underlie the learning task itself and remain consistent across models, regardless of scale. From this framework, we propose the \textbf{Linear Representation Transferability (LRT)} Hypothesis -- that there exists an affine transformation between the representation spaces of different models. To test this hypothesis, we learn affine mappings between the hidden states of models of different sizes and evaluate whether steering vectors -- directions in hidden state space associated with specific model behaviors -- retain their semantic effect when transferred from small to large language models using the learned mappings. We find strong empirical evidence that such affine mappings can preserve steering behaviors. These findings suggest that representations learned by small models can be used to guide the behavior of large models, and that the LRT hypothesis may be a promising direction on understanding representation alignment across model scales.
- [79] arXiv:2506.00656 [pdf, html, other]
-
Title: Permutation-Invariant Transformer Neural Architectures for Set-Based Indoor Localization Using Learned RSSI EmbeddingsComments: 16 pages, 1 figureSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
We propose a permutation-invariant neural architecture for indoor localization using RSSI scans from Wi-Fi access points. Each scan is modeled as an unordered set of (BSSID, RSSI) pairs, where BSSIDs are mapped to learned embeddings and concatenated with signal strength. These are processed by a Set Transformer, enabling the model to handle variable-length, sparse inputs while learning attention- based representations over access point relationships. We evaluate the model on a dataset collected across a campus environment consisting of six buildings. Results show that the model accurately recovers fine-grained spatial structure and maintains performance across physically distinct domains. In our experiments, a simple LSTM consistently outperformed all other models, achieving the lowest mean localization error across three tasks (E1 - E3), with average errors as low as 2.23 m. The Set Transformer performed competitively, ranking second in every experiment and outperforming the MLP, RNN, and basic attention models, particularly in scenarios involving multiple buildings (E2) and multiple floors (E3). Performance degraded most in E2, where signal conditions varied substantially across buildings, highlighting the importance of architectural robustness to domain diversity. This work demonstrates that set-based neural models are a natural fit for signal-based localization, offering a principled approach to handling sparse, unordered inputs in real-world positioning tasks.
- [80] arXiv:2506.00660 [pdf, html, other]
-
Title: Differential Privacy for Deep Learning in MedicineMarziyeh Mohammadi, Mohsen Vejdanihemmat, Mahshad Lotfinia, Mirabela Rusu, Daniel Truhn, Andreas Maier, Soroosh Tayebi ArastehSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Differential privacy (DP) is a key technique for protecting sensitive patient data in medical deep learning (DL). As clinical models grow more data-dependent, balancing privacy with utility and fairness has become a critical challenge. This scoping review synthesizes recent developments in applying DP to medical DL, with a particular focus on DP-SGD and alternative mechanisms across centralized and federated settings. Using a structured search strategy, we identified 74 studies published up to March 2025. Our analysis spans diverse data modalities, training setups, and downstream tasks, and highlights the tradeoffs between privacy guarantees, model accuracy, and subgroup fairness. We find that while DP-especially at strong privacy budgets-can preserve performance in well-structured imaging tasks, severe degradation often occurs under strict privacy, particularly in underrepresented or complex modalities. Furthermore, privacy-induced performance gaps disproportionately affect demographic subgroups, with fairness impacts varying by data type and task. A small subset of studies explicitly addresses these tradeoffs through subgroup analysis or fairness metrics, but most omit them entirely. Beyond DP-SGD, emerging approaches leverage alternative mechanisms, generative models, and hybrid federated designs, though reporting remains inconsistent. We conclude by outlining key gaps in fairness auditing, standardization, and evaluation protocols, offering guidance for future work toward equitable and clinically robust privacy-preserving DL systems in medicine.
- [81] arXiv:2506.00676 [pdf, html, other]
-
Title: SafeTuneBed: A Toolkit for Benchmarking LLM Safety Alignment in Fine-TuningSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
As large language models (LLMs) become ubiquitous, parameter-efficient fine-tuning methods and safety-first defenses have proliferated rapidly. However, the number of approaches and their recent increase have resulted in diverse evaluations-varied datasets, metrics, and inconsistent threat settings-making it difficult to fairly compare safety, utility, and robustness across methods. To address this, we introduce SafeTuneBed, a benchmark and toolkit unifying fine-tuning and defense evaluation. SafeTuneBed (i) curates a diverse repository of multiple fine-tuning datasets spanning sentiment analysis, question-answering, multi-step reasoning, and open-ended instruction tasks, and allows for the generation of harmful-variant splits; (ii) enables integration of state-of-the-art defenses, including alignment-stage immunization, in-training safeguards, and post-tuning repair; and (iii) provides evaluators for safety (attack success rate, refusal consistency) and utility. Built on Python-first, dataclass-driven configs and plugins, SafeTuneBed requires minimal additional code to specify any fine-tuning regime, defense method, and metric suite, while ensuring end-to-end reproducibility. We showcase its value by benchmarking representative defenses across varied poisoning scenarios and tasks. By standardizing data, code, and metrics, SafeTuneBed is the first focused toolkit of its kind to accelerate rigorous and comparable research in safe LLM fine-tuning. Code is available at: this https URL
- [82] arXiv:2506.00688 [pdf, html, other]
-
Title: Existing Large Language Model Unlearning Evaluations Are InconclusiveZhili Feng, Yixuan Even Xu, Alexander Robey, Robert Kirk, Xander Davies, Yarin Gal, Avi Schwarzschild, J. Zico KolterSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Machine unlearning aims to remove sensitive or undesired data from large language models. However, recent studies suggest that unlearning is often shallow, claiming that removed knowledge can easily be recovered. In this work, we critically examine standard unlearning evaluation practices and uncover key limitations that shake our trust in those findings. First, we show that some evaluations introduce substantial new information into the model, potentially masking true unlearning performance by re-teaching the model during testing. Second, we demonstrate that evaluation outcomes vary significantly across tasks, undermining the generalizability of current evaluation routines. Finally, we find that many evaluations rely on spurious correlations, making their results difficult to trust and interpret. Taken together, these issues suggest that current evaluation protocols may both overstate and understate unlearning success. To address this, we propose two principles for future unlearning evaluations: minimal information injection and downstream task awareness. We validate these principles through a series of targeted experiments, showing how violations of each can lead to misleading conclusions.
- [83] arXiv:2506.00691 [pdf, html, other]
-
Title: Optimizing Sensory Neurons: Nonlinear Attention Mechanisms for Accelerated Convergence in Permutation-Invariant Neural Networks for Reinforcement LearningSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Training reinforcement learning (RL) agents often requires significant computational resources and extended training times. To address this, we build upon the foundation laid by Google Brain's Sensory Neuron, which introduced a novel neural architecture for reinforcement learning tasks that maintained permutation in-variance in the sensory neuron system. While the baseline model demonstrated significant performance improvements over traditional approaches, we identified opportunities to enhance the efficiency of the learning process further. We propose a modified attention mechanism incorporating a non-linear transformation of the key vectors (K) using a mapping function, resulting in a new set of key vectors (K'). This non-linear mapping enhances the representational capacity of the attention mechanism, allowing the model to encode more complex feature interactions and accelerating convergence without compromising performance. Our enhanced model demonstrates significant improvements in learning efficiency, showcasing the potential for non-linear attention mechanisms in advancing reinforcement learning algorithms.
- [84] arXiv:2506.00700 [pdf, html, other]
-
Title: Central Path Proximal Policy OptimizationSubjects: Machine Learning (cs.LG)
In constrained Markov decision processes, enforcing constraints during training is often thought of as decreasing the final return. Recently, it was shown that constraints can be incorporated directly in the policy geometry, yielding an optimization trajectory close to the central path of a barrier method, which does not compromise final return. Building on this idea, we introduce Central Path Proximal Policy Optimization (C3PO), a simple modification of PPO that produces policy iterates, which stay close to the central path of the constrained optimization problem. Compared to existing on-policy methods, C3PO delivers improved performance with tighter constraint enforcement, suggesting that central path-guided updates offer a promising direction for constrained policy optimization.
- [85] arXiv:2506.00701 [pdf, html, other]
-
Title: Bayesian Inference of Training Dataset MembershipComments: 10 pagesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Determining whether a dataset was part of a machine learning model's training data pool can reveal privacy vulnerabilities, a challenge often addressed through membership inference attacks (MIAs). Traditional MIAs typically require access to model internals or rely on computationally intensive shadow models. This paper proposes an efficient, interpretable and principled Bayesian inference method for membership inference. By analyzing post-hoc metrics such as prediction error, confidence (entropy), perturbation magnitude, and dataset statistics from a trained ML model, our approach computes posterior probabilities of membership without requiring extensive model training. Experimental results on synthetic datasets demonstrate the method's effectiveness in distinguishing member from non-member datasets. Beyond membership inference, this method can also detect distribution shifts, offering a practical and interpretable alternative to existing approaches.
- [86] arXiv:2506.00710 [pdf, html, other]
-
Title: RelDiff: Relational Data Generative Modeling with Graph-Based Diffusion ModelsValter Hudovernik, Minkai Xu, Juntong Shi, Lovro Šubelj, Stefano Ermon, Erik Štrumbelj, Jure LeskovecSubjects: Machine Learning (cs.LG)
Real-world databases are predominantly relational, comprising multiple interlinked tables that contain complex structural and statistical dependencies. Learning generative models on relational data has shown great promise in generating synthetic data and imputing missing values. However, existing methods often struggle to capture this complexity, typically reducing relational data to conditionally generated flat tables and imposing limiting structural assumptions. To address these limitations, we introduce RelDiff, a novel diffusion generative model that synthesizes complete relational databases by explicitly modeling their foreign key graph structure. RelDiff combines a joint graph-conditioned diffusion process across all tables for attribute synthesis, and a $2K+$SBM graph generator based on the Stochastic Block Model for structure generation. The decomposition of graph structure and relational attributes ensures both high fidelity and referential integrity, both of which are crucial aspects of synthetic relational database generation. Experiments on 11 benchmark datasets demonstrate that RelDiff consistently outperforms prior methods in producing realistic and coherent synthetic relational databases. Code is available at this https URL.
- [87] arXiv:2506.00711 [pdf, html, other]
-
Title: QoQ-Med: Building Multimodal Clinical Foundation Models with Domain-Aware GRPO TrainingSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
Clinical decision-making routinely demands reasoning over heterogeneous data, yet existing multimodal language models (MLLMs) remain largely vision-centric and fail to generalize across clinical specialties. To bridge this gap, we introduce QoQ-Med-7B/32B, the first open generalist clinical foundation model that jointly reasons across medical images, time-series signals, and text reports. QoQ-Med is trained with Domain-aware Relative Policy Optimization (DRPO), a novel reinforcement-learning objective that hierarchically scales normalized rewards according to domain rarity and modality difficulty, mitigating performance imbalance caused by skewed clinical data distributions. Trained on 2.61 million instruction tuning pairs spanning 9 clinical domains, we show that DRPO training boosts diagnostic performance by 43% in macro-F1 on average across all visual domains as compared to other critic-free training methods like GRPO. Furthermore, with QoQ-Med trained on intensive segmentation data, it is able to highlight salient regions related to the diagnosis, with an IoU 10x higher than open models while reaching the performance of OpenAI o4-mini. To foster reproducibility and downstream research, we release (i) the full model weights, (ii) the modular training pipeline, and (iii) all intermediate reasoning traces at this https URL.
- [88] arXiv:2506.00723 [pdf, html, other]
-
Title: Pitfalls in Evaluating Language Model ForecastersComments: 20 pages, 8 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
Large language models (LLMs) have recently been applied to forecasting tasks, with some works claiming these systems match or exceed human performance. In this paper, we argue that, as a community, we should be careful about such conclusions as evaluating LLM forecasters presents unique challenges. We identify two broad categories of issues: (1) difficulty in trusting evaluation results due to many forms of temporal leakage, and (2) difficulty in extrapolating from evaluation performance to real-world forecasting. Through systematic analysis and concrete examples from prior work, we demonstrate how evaluation flaws can raise concerns about current and future performance claims. We argue that more rigorous evaluation methodologies are needed to confidently assess the forecasting abilities of LLMs.
- [89] arXiv:2506.00724 [pdf, html, other]
-
Title: A condensing approach to multiple shooting neural ordinary differential equationSubjects: Machine Learning (cs.LG); Dynamical Systems (math.DS)
Multiple-shooting is a parameter estimation approach for ordinary differential equations. In this approach, the trajectory is broken into small intervals, each of which can be integrated independently. Equality constraints are then applied to eliminate the shooting gap between the end of the previous trajectory and the start of the next trajectory. Unlike single-shooting, multiple-shooting is more stable, especially for highly oscillatory and long trajectories. In the context of neural ordinary differential equations, multiple-shooting is not widely used due to the challenge of incorporating general equality constraints. In this work, we propose a condensing-based approach to incorporate these shooting equality constraints while training a multiple-shooting neural ordinary differential equation (MS-NODE) using first-order optimization methods such as Adam.
- [90] arXiv:2506.00727 [pdf, html, other]
-
Title: Adaptive Plane Reformatting for 4D Flow MRI using Deep Reinforcement LearningJavier Bisbal, Julio Sotelo, Maria I Valdés, Pablo Irarrazaval, Marcelo E Andia, Julio García, José Rodriguez-Palomarez, Francesca Raimondi, Cristián Tejos, Sergio UribeComments: 11 pages, 4 figures, submitted to IEEE Transactions on Medical ImagingSubjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
Deep reinforcement learning (DRL) algorithms have shown robust results in plane reformatting tasks. In these methods, an agent sequentially adjusts the position and orientation of an initial plane towards an objective location. This process allows accurate plane reformatting, without the need for detailed landmarks, which makes it suitable for images with limited contrast and resolution, such as 4D flow MRI. However, current DRL methods require the test dataset to be in the same position and orientation as the training dataset. In this paper, we present a novel technique that utilizes a flexible coordinate system based on the current state, enabling navigation in volumes at any position or orientation. We adopted the Asynchronous Advantage Actor Critic (A3C) algorithm for reinforcement learning, outperforming Deep Q Network (DQN). Experimental results in 4D flow MRI demonstrate improved accuracy in plane reformatting angular and distance errors (6.32 +- 4.15 ° and 3.40 +- 2.75 mm), as well as statistically equivalent flow measurements determined by a plane reformatting process done by an expert (p=0.21). The method's flexibility and adaptability make it a promising candidate for other medical imaging applications beyond 4D flow MRI.
- [91] arXiv:2506.00731 [pdf, html, other]
-
Title: MoPINNEnKF: Iterative Model Inference using generic-PINN-based ensemble Kalman filterSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Physics-informed neural networks (PINNs) have emerged as a powerful tool for solving forward and inverse problems involving partial differential equations (PDEs) by incorporating physical laws into the training process. However, the performance of PINNs is often hindered in real-world scenarios involving noisy observational data and missing physics, particularly in inverse problems. In this work, we propose an iterative multi-objective PINN ensemble Kalman filter (MoPINNEnKF) framework that improves the robustness and accuracy of PINNs in both forward and inverse problems by using the \textit{ensemble Kalman filter} and the \textit{non-dominated sorting genetic algorithm} III (NSGA-III). Specifically, NSGA-III is used as a multi-objective optimizer that can generate various ensemble members of PINNs along the optimal Pareto front, while accounting the model uncertainty in the solution space. These ensemble members are then utilized within the EnKF to assimilate noisy observational data. The EnKF's analysis is subsequently used to refine the data loss component for retraining the PINNs, thereby iteratively updating their parameters. The iterative procedure generates improved solutions to the PDEs. The proposed method is tested on two benchmark problems: the one-dimensional viscous Burgers equation and the time-fractional mixed diffusion-wave equation (TFMDWE). The numerical results show it outperforms standard PINNs in handling noisy data and missing physics.
- [92] arXiv:2506.00732 [pdf, html, other]
-
Title: Bregman Conditional Random Fields: Sequence Labeling with Parallelizable Inference AlgorithmsComments: ACL 2025Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
We propose a novel discriminative model for sequence labeling called Bregman conditional random fields (BCRF). Contrary to standard linear-chain conditional random fields, BCRF allows fast parallelizable inference algorithms based on iterative Bregman projections. We show how such models can be learned using Fenchel-Young losses, including extension for learning from partial labels. Experimentally, our approach delivers comparable results to CRF while being faster, and achieves better results in highly constrained settings compared to mean field, another parallelizable alternative.
- [93] arXiv:2506.00744 [pdf, html, other]
-
Title: Blending Complementary Memory Systems in Hybrid Quadratic-Linear TransformersSubjects: Machine Learning (cs.LG)
We develop hybrid memory architectures for general-purpose sequence processing neural networks, that combine key-value memory using softmax attention (KV-memory) with dynamic synaptic memory through fast-weight programming (FW-memory) -- the core principles of quadratic and linear transformers, respectively. These two memory systems have complementary but individually limited properties: KV-memory offers precise retrieval but is constrained by quadratic complexity in sequence length, while FW-memory supports arbitrarily long sequences and enables more expressive computation but sacrifices precise recall. We propose and compare three methods to blend these two systems into a single memory system to leverage the strengths of both. We conduct experiments on general language modeling and retrieval tasks by training 340M- and 1.3B-parameter models from scratch, as well as on synthetic algorithmic tasks designed to precisely illustrate the benefits of certain hybrid methods over others. We also evaluate our hybrid memory systems on reinforcement learning in partially observable environments. Overall, we demonstrate how a well-designed hybrid can overcome the limitations of its individual components, offering new insights into the design principle of neural memory systems.
- [94] arXiv:2506.00756 [pdf, html, other]
-
Title: "Who experiences large model decay and why?" A Hierarchical Framework for Diagnosing Heterogeneous Performance DriftComments: 13 pages, 9 figures, 8 tables, 18 pages appendix. To be published in Proceedings of the 42nd International Conference on Machine Learning, Vancouver, Canada. PMLR 267, 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Machine learning (ML) models frequently experience performance degradation when deployed in new contexts. Such degradation is rarely uniform: some subgroups may suffer large performance decay while others may not. Understanding where and how large differences in performance arise is critical for designing targeted corrective actions that mitigate decay for the most affected subgroups while minimizing any unintended effects. Current approaches do not provide such detailed insight, as they either (i) explain how average performance shifts arise or (ii) identify adversely affected subgroups without insight into how this occurred. To this end, we introduce a Subgroup-scanning Hierarchical Inference Framework for performance drifT (SHIFT). SHIFT first asks "Is there any subgroup with unacceptably large performance decay due to covariate/outcome shifts?" (Where?) and, if so, dives deeper to ask "Can we explain this using more detailed variable(subset)-specific shifts?" (How?). In real-world experiments, we find that SHIFT identifies interpretable subgroups affected by performance decay, and suggests targeted actions that effectively mitigate the decay.
- [95] arXiv:2506.00764 [pdf, html, other]
-
Title: Learning Juntas under Markov Random FieldsSubjects: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS)
We give an algorithm for learning $O(\log n)$ juntas in polynomial-time with respect to Markov Random Fields (MRFs) in a smoothed analysis framework where only the external field has been randomly perturbed. This is a broad generalization of the work of Kalai and Teng, who gave an algorithm that succeeded with respect to smoothed product distributions (i.e., MRFs whose dependency graph has no edges). Our algorithm has two phases: (1) an unsupervised structure learning phase and (2) a greedy supervised learning algorithm. This is the first example where algorithms for learning the structure of an undirected graphical model lead to provably efficient algorithms for supervised learning.
- [96] arXiv:2506.00770 [pdf, html, other]
-
Title: Beyond Attention: Learning Spatio-Temporal Dynamics with Emergent Interpretable TopologiesComments: 13 pages, 10 figures, workshopSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Social and Information Networks (cs.SI)
Spatio-temporal forecasting is critical in applications such as traffic prediction, energy demand modeling, and weather monitoring. While Graph Attention Networks (GATs) are popular for modeling spatial dependencies, they rely on predefined adjacency structures and dynamic attention scores, introducing inductive biases and computational overhead that can obscure interpretability.
We propose InterGAT, a simplified alternative to GAT that replaces masked attention with a fully learnable, symmetric node interaction matrix, capturing latent spatial relationships without relying on fixed graph topologies. Our framework, InterGAT-GRU, which incorporates a GRU-based temporal decoder, outperforms the baseline GAT-GRU in forecasting accuracy, achieving at least a 21% improvement on the SZ-Taxi dataset and a 6% improvement on the Los-Loop dataset across all forecasting horizons (15 to 60 minutes). Additionally, we observed reduction in training time by 60-70% compared to GAT-GRU baseline.
Crucially, the learned interaction matrix reveals interpretable structure: it recovers sparse, topology-aware attention patterns that align with community structure. Spectral and clustering analyses show that the model captures both localized and global dynamics, offering insights into the functional topology driving predictions. This highlights how structure learning can simultaneously support prediction, computational efficiency, and topological interpretabil-ity in dynamic graph-based domains. - [97] arXiv:2506.00771 [pdf, html, other]
-
Title: Manipulating 3D Molecules in a Fixed-Dimensional SE(3)-Equivariant Latent SpaceComments: 11 pages, 6 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Medicinal chemists often optimize drugs considering their 3D structures and designing structurally distinct molecules that retain key features, such as shapes, pharmacophores, or chemical properties. Previous deep learning approaches address this through supervised tasks like molecule inpainting or property-guided optimization. In this work, we propose a flexible zero-shot molecule manipulation method by navigating in a shared latent space of 3D molecules. We introduce a Variational AutoEncoder (VAE) for 3D molecules, named MolFLAE, which learns a fixed-dimensional, SE(3)-equivariant latent space independent of atom counts. MolFLAE encodes 3D molecules using an SE(3)-equivariant neural network into fixed number of latent nodes, distinguished by learned embeddings. The latent space is regularized, and molecular structures are reconstructed via a Bayesian Flow Network (BFN) conditioned on the encoder's latent output. MolFLAE achieves competitive performance on standard unconditional 3D molecule generation benchmarks. Moreover, the latent space of MolFLAE enables zero-shot molecule manipulation, including atom number editing, structure reconstruction, and coordinated latent interpolation for both structure and properties. We further demonstrate our approach on a drug optimization task for the human glucocorticoid receptor, generating molecules with improved hydrophilicity while preserving key interactions, under computational evaluations. These results highlight the flexibility, robustness, and real-world utility of our method, opening new avenues for molecule editing and optimization.
- [98] arXiv:2506.00772 [pdf, html, other]
-
Title: LIFT the Veil for the Truth: Principal Weights Emerge after Rank Reduction for Reasoning-Focused Supervised Fine-TuningZihang Liu, Tianyu Pang, Oleg Balabanov, Chaoqun Yang, Tianjin Huang, Lu Yin, Yaoqing Yang, Shiwei LiuComments: ICML 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Recent studies have shown that supervised fine-tuning of LLMs on a small number of high-quality datasets can yield strong reasoning capabilities. However, full fine-tuning (Full FT), while powerful, is computationally expensive and susceptible to overfitting and catastrophic forgetting, particularly when data is limited. Sparse fine-tuning, which previously achieved notable success by updating only a small subset of model parameters, offers a promising trade-off between efficiency and effectiveness. Yet, it has lagged behind in the LLM era due to the difficulty of identifying parameters truly critical for reasoning. In this work, we state that weights with the largest magnitude after low-rank approximation are critical weights for fine-tuning, which we call Principal Weights. Surprisingly, while magnitude-based sparse fine-tuning performs poorly as a baseline on LLM fine-tuning, it becomes highly effective after rank reduction. These insights motivate our method: Low-rank Informed Sparse Fine-Tuning (LIFT). LIFT only updates the top 5% Principal Weights throughout training and consistently achieves better performance on reasoning tasks than Full FT, while maintaining memory efficiency on par with popular parameter-efficient fine-tuning methods. In addition to strong performance on target domains such as arithmetic reasoning, LIFT also retains up to 20% more source-domain knowledge, compared to Full FT and LoRA. Our code is available at: this https URL.
- [99] arXiv:2506.00795 [pdf, html, other]
-
Title: Bridging Supervised and Temporal Difference Learning with $Q$-Conditioned MaximizationSubjects: Machine Learning (cs.LG)
Recently, supervised learning (SL) methodology has emerged as an effective approach for offline reinforcement learning (RL) due to their simplicity, stability, and efficiency. However, recent studies show that SL methods lack the trajectory stitching capability, typically associated with temporal difference (TD)-based approaches. A question naturally surfaces: How can we endow SL methods with stitching capability and bridge its performance gap with TD learning? To answer this question, we introduce $Q$-conditioned maximization supervised learning for offline goal-conditioned RL, which enhances SL with the stitching capability through $Q$-conditioned policy and $Q$-conditioned maximization. Concretely, we propose Goal-Conditioned Reinforced Supervised Learning (GCReinSL), which consists of (1) estimating the $Q$-function by CVAE from the offline dataset and (2) finding the maximum $Q$-value within the data support by integrating $Q$-function maximization with Expectile Regression. In inference time, our policy chooses optimal actions based on such a maximum $Q$-value. Experimental results from stitching evaluations on offline RL datasets demonstrate that our method outperforms prior SL approaches with stitching capabilities and goal data augmentation techniques.
- [100] arXiv:2506.00797 [pdf, html, other]
-
Title: Action Dependency Graphs for Globally Optimal Coordinated Reinforcement LearningSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Systems and Control (eess.SY); Optimization and Control (math.OC)
Action-dependent individual policies, which incorporate both environmental states and the actions of other agents in decision-making, have emerged as a promising paradigm for achieving global optimality in multi-agent reinforcement learning (MARL). However, the existing literature often adopts auto-regressive action-dependent policies, where each agent's policy depends on the actions of all preceding agents. This formulation incurs substantial computational complexity as the number of agents increases, thereby limiting scalability. In this work, we consider a more generalized class of action-dependent policies, which do not necessarily follow the auto-regressive form. We propose to use the `action dependency graph (ADG)' to model the inter-agent action dependencies. Within the context of MARL problems structured by coordination graphs, we prove that an action-dependent policy with a sparse ADG can achieve global optimality, provided the ADG satisfies specific conditions specified by the coordination graph. Building on this theoretical foundation, we develop a tabular policy iteration algorithm with guaranteed global optimality. Furthermore, we integrate our framework into several SOTA algorithms and conduct experiments in complex environments. The empirical results affirm the robustness and applicability of our approach in more general scenarios, underscoring its potential for broader MARL challenges.
- [101] arXiv:2506.00798 [pdf, html, other]
-
Title: A Dynamic Stiefel Graph Neural Network for Efficient Spatio-Temporal Time Series ForecastingComments: Accepted at IJCAI 2025Subjects: Machine Learning (cs.LG)
Spatio-temporal time series (STTS) have been widely used in many applications. However, accurately forecasting STTS is challenging due to complex dynamic correlations in both time and space dimensions. Existing graph neural networks struggle to balance effectiveness and efficiency in modeling dynamic spatio-temporal relations. To address this problem, we propose the Dynamic Spatio-Temporal Stiefel Graph Neural Network (DST-SGNN) to efficiently process STTS. For DST-SGNN, we first introduce the novel Stiefel Graph Spectral Convolution (SGSC) and Stiefel Graph Fourier Transform (SGFT). The SGFT matrix in SGSC is constrained to lie on the Stiefel manifold, and SGSC can be regarded as a filtered graph spectral convolution. We also propose the Linear Dynamic Graph Optimization on Stiefel Manifold (LDGOSM), which can efficiently learn the SGFT matrix from the dynamic graph and significantly reduce the computational complexity. Finally, we propose a multi-layer SGSC (MSGSC) that efficiently captures complex spatio-temporal correlations. Extensive experiments on seven spatio-temporal datasets show that DST-SGNN outperforms state-of-the-art methods while maintaining relatively low computational costs.
- [102] arXiv:2506.00799 [pdf, html, other]
-
Title: Uni-LoRA: One Vector is All You NeedSubjects: Machine Learning (cs.LG)
Low-Rank Adaptation (LoRA) has become the de facto parameter-efficient fine-tuning (PEFT) method for large language models (LLMs) by constraining weight updates to low-rank matrices. Recent works such as Tied-LoRA, VeRA, and VB-LoRA push efficiency further by introducing additional constraints to reduce the trainable parameter space. In this paper, we show that the parameter space reduction strategies employed by these LoRA variants can be formulated within a unified framework, Uni-LoRA, where the LoRA parameter space, flattened as a high-dimensional vector space $R^D$, can be reconstructed through a projection from a subspace R^d, with $d \ll D$. We demonstrate that the fundamental difference among various LoRA methods lies in the choice of the projection matrix, $P \in R^{D \times d}$.Most existing LoRA variants rely on layer-wise or structure-specific projections that limit cross-layer parameter sharing, thereby compromising parameter efficiency. In light of this, we introduce an efficient and theoretically grounded projection matrix that is isometric, enabling global parameter sharing and reducing computation overhead. Furthermore, under the unified view of Uni-LoRA, this design requires only a single trainable vector to reconstruct LoRA parameters for the entire LLM - making Uni-LoRA both a unified framework and a "one-vector-only" solution. Extensive experiments on GLUE, mathematical reasoning, and instruction tuning benchmarks demonstrate that Uni-LoRA achieves state-of-the-art parameter efficiency while outperforming or matching prior approaches in predictive performance.
- [103] arXiv:2506.00808 [pdf, html, other]
-
Title: Unlearning Inversion Attacks for Graph Neural NetworksSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
Graph unlearning methods aim to efficiently remove the impact of sensitive data from trained GNNs without full retraining, assuming that deleted information cannot be recovered. In this work, we challenge this assumption by introducing the graph unlearning inversion attack: given only black-box access to an unlearned GNN and partial graph knowledge, can an adversary reconstruct the removed edges? We identify two key challenges: varying probability-similarity thresholds for unlearned versus retained edges, and the difficulty of locating unlearned edge endpoints, and address them with TrendAttack. First, we derive and exploit the confidence pitfall, a theoretical and empirical pattern showing that nodes adjacent to unlearned edges exhibit a large drop in model confidence. Second, we design an adaptive prediction mechanism that applies different similarity thresholds to unlearned and other membership edges. Our framework flexibly integrates existing membership inference techniques and extends them with trend features. Experiments on four real-world datasets demonstrate that TrendAttack significantly outperforms state-of-the-art GNN membership inference baselines, exposing a critical privacy vulnerability in current graph unlearning methods.
- [104] arXiv:2506.00844 [pdf, html, other]
-
Title: LLM Cannot Discover Causality, and Should Be Restricted to Non-Decisional Support in Causal DiscoverySubjects: Machine Learning (cs.LG)
This paper critically re-evaluates LLMs' role in causal discovery and argues against their direct involvement in determining causal relationships. We demonstrate that LLMs' autoregressive, correlation-driven modeling inherently lacks the theoretical grounding for causal reasoning and introduces unreliability when used as priors in causal discovery algorithms. Through empirical studies, we expose the limitations of existing LLM-based methods and reveal that deliberate prompt engineering (e.g., injecting ground-truth knowledge) could overstate their performance, helping to explain the consistently favorable results reported in much of the current literature. Based on these findings, we strictly confined LLMs' role to a non-decisional auxiliary capacity: LLMs should not participate in determining the existence or directionality of causal relationships, but can assist the search process for causal graphs (e.g., LLM-based heuristic search). Experiments across various settings confirm that, by strictly isolating LLMs from causal decision-making, LLM-guided heuristic search can accelerate the convergence and outperform both traditional and LLM-based methods in causal structure learning. We conclude with a call for the community to shift focus from naively applying LLMs to developing specialized models and training method that respect the core principles of causal discovery.
- [105] arXiv:2506.00845 [pdf, html, other]
-
Title: Generalizable LLM Learning of Graph Synthetic Data with Reinforcement LearningComments: 9 pages, 3 figures, 3 tables. Experimental code and results are publicly available at this https URLSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Previous research has sought to enhance the graph reasoning capabilities of LLMs by supervised fine-tuning on synthetic graph data. While these led to specialized LLMs better at solving graph algorithm problems, we don't need LLMs for shortest path: we need generalization from synthetic graph data to real-world tasks with implicit graph structures. In this work, we propose to unlock generalizable learning of graph synthetic data with reinforcement learning. We first design solution-based and process-based rewards for synthetic graph problems: instead of rigid memorizing response patterns in direct fine-tuning, we posit that RL would help LLMs grasp the essentials underlying graph reasoning and alleviate overfitting. We employ RL algorithms such as GRPO and DPO, aligning both off-the-shelf LLMs and LLMs fine-tuned on synthetic graph data. We then compare them against existing settings on both in-domain synthetic tasks and out-of-domain real-world tasks with implicit graph structures such as multi-hop QA, structured planning, and more. Extensive experiments demonstrate that our RL recipe leads to statistically significant improvement on 5 datasets, with an average gain of 12.9\% over baseline settings. Further analysis reveals that process-based rewards consistently outperform solution-based rewards, mixing synthetic and real-world task data yields potential gains, while compositionality and explainable intermediate steps remains a critical challenge even after RL.
- [106] arXiv:2506.00846 [pdf, html, other]
-
Title: Infinite-Width Limit of a Single Attention Layer: Analysis via Tensor ProgramsSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
In modern theoretical analyses of neural networks, the infinite-width limit is often invoked to justify Gaussian approximations of neuron preactivations (e.g., via neural network Gaussian processes or Tensor Programs). However, these Gaussian-based asymptotic theories have so far been unable to capture the behavior of attention layers, except under special regimes such as infinitely many heads or tailored scaling schemes. In this paper, leveraging the Tensor Programs framework, we rigorously identify the infinite-width limit distribution of variables within a single attention layer under realistic architectural dimensionality and standard $1/\sqrt{n}$-scaling with $n$ dimensionality. We derive the exact form of this limit law without resorting to infinite-head approximations or tailored scalings, demonstrating that it departs fundamentally from Gaussianity. This limiting distribution exhibits non-Gaussianity from a hierarchical structure, being Gaussian conditional on the random similarity scores. Numerical experiments validate our theoretical predictions, confirming the effectiveness of our theory at finite width and accurate description of finite-head attentions. Beyond characterizing a standalone attention layer, our findings lay the groundwork for developing a unified theory of deep Transformer architectures in the infinite-width regime.
- [107] arXiv:2506.00848 [pdf, html, other]
-
Title: Speech UnlearningComments: Interspeech 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Sound (cs.SD); Audio and Speech Processing (eess.AS)
We introduce machine unlearning for speech tasks, a novel and underexplored research problem that aims to efficiently and effectively remove the influence of specific data from trained speech models without full retraining. This has important applications in privacy preservation, removal of outdated or noisy data, and bias mitigation. While machine unlearning has been studied in computer vision and natural language processing, its application to speech is largely unexplored due to the high-dimensional, sequential, and speaker-dependent nature of speech data. We define two fundamental speech unlearning tasks: sample unlearning, which removes individual data points (e.g., a voice recording), and class unlearning, which removes an entire category (e.g., all data from a speaker), while preserving performance on the remaining data. Experiments on keyword spotting and speaker identification demonstrate that unlearning speech data is significantly more challenging than unlearning image or text data. We conclude with key future directions in this area, including structured training, robust evaluation, feature-level unlearning, broader applications, scalable methods, and adversarial robustness.
- [108] arXiv:2506.00849 [pdf, html, other]
-
Title: Generalization in VAE and Diffusion Models: A Unified Information-Theoretic AnalysisComments: ICLR 2025 AcceptedSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Despite the empirical success of Diffusion Models (DMs) and Variational Autoencoders (VAEs), their generalization performance remains theoretically underexplored, especially lacking a full consideration of the shared encoder-generator structure. Leveraging recent information-theoretic tools, we propose a unified theoretical framework that provides guarantees for the generalization of both the encoder and generator by treating them as randomized mappings. This framework further enables (1) a refined analysis for VAEs, accounting for the generator's generalization, which was previously overlooked; (2) illustrating an explicit trade-off in generalization terms for DMs that depends on the diffusion time $T$; and (3) providing computable bounds for DMs based solely on the training data, allowing the selection of the optimal $T$ and the integration of such bounds into the optimization process to improve model performance. Empirical results on both synthetic and real datasets illustrate the validity of the proposed theory.
- [109] arXiv:2506.00862 [pdf, html, other]
-
Title: FourierFlow: Frequency-aware Flow Matching for Generative Turbulence ModelingComments: 27 pages, 14 figuresSubjects: Machine Learning (cs.LG)
Modeling complex fluid systems, especially turbulence governed by partial differential equations (PDEs), remains a fundamental challenge in science and engineering. Recently, diffusion-based generative models have gained attention as a powerful approach for these tasks, owing to their capacity to capture long-range dependencies and recover hierarchical structures. However, we present both empirical and theoretical evidence showing that generative models struggle with significant spectral bias and common-mode noise when generating high-fidelity turbulent flows. Here we propose FourierFlow, a novel generative turbulence modeling framework that enhances the frequency-aware learning by both implicitly and explicitly mitigating spectral bias and common-mode noise. FourierFlow comprises three key innovations. Firstly, we adopt a dual-branch backbone architecture, consisting of a salient flow attention branch with local-global awareness to focus on sensitive turbulence areas. Secondly, we introduce a frequency-guided Fourier mixing branch, which is integrated via an adaptive fusion strategy to explicitly mitigate spectral bias in the generative model. Thirdly, we leverage the high-frequency modeling capabilities of the masked auto-encoder pre-training and implicitly align the features of the generative model toward high-frequency components. We validate the effectiveness of FourierFlow on three canonical turbulent flow scenarios, demonstrating superior performance compared to state-of-the-art methods. Furthermore, we show that our model exhibits strong generalization capabilities in challenging settings such as out-of-distribution domains, long-term temporal extrapolation, and robustness to noisy inputs. The code can be found at this https URL.
- [110] arXiv:2506.00867 [pdf, html, other]
-
Title: Local Manifold Approximation and Projection for Manifold-Aware Diffusion PlanningComments: ICML 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Recent advances in diffusion-based generative modeling have demonstrated significant promise in tackling long-horizon, sparse-reward tasks by leveraging offline datasets. While these approaches have achieved promising results, their reliability remains inconsistent due to the inherent stochastic risk of producing infeasible trajectories, limiting their applicability in safety-critical applications. We identify that the primary cause of these failures is inaccurate guidance during the sampling procedure, and demonstrate the existence of manifold deviation by deriving a lower bound on the guidance gap. To address this challenge, we propose Local Manifold Approximation and Projection (LoMAP), a training-free method that projects the guided sample onto a low-rank subspace approximated from offline datasets, preventing infeasible trajectory generation. We validate our approach on standard offline reinforcement learning benchmarks that involve challenging long-horizon planning. Furthermore, we show that, as a standalone module, LoMAP can be incorporated into the hierarchical diffusion planner, providing further performance enhancements.
- [111] arXiv:2506.00880 [pdf, html, other]
-
Title: ModuLM: Enabling Modular and Multimodal Molecular Relational Learning with Large Language ModelsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Quantitative Methods (q-bio.QM)
Molecular Relational Learning (MRL) aims to understand interactions between molecular pairs, playing a critical role in advancing biochemical research. With the recent development of large language models (LLMs), a growing number of studies have explored the integration of MRL with LLMs and achieved promising results. However, the increasing availability of diverse LLMs and molecular structure encoders has significantly expanded the model space, presenting major challenges for benchmarking. Currently, there is no LLM framework that supports both flexible molecular input formats and dynamic architectural switching. To address these challenges, reduce redundant coding, and ensure fair model comparison, we propose ModuLM, a framework designed to support flexible LLM-based model construction and diverse molecular representations. ModuLM provides a rich suite of modular components, including 8 types of 2D molecular graph encoders, 11 types of 3D molecular conformation encoders, 7 types of interaction layers, and 7 mainstream LLM backbones. Owing to its highly flexible model assembly mechanism, ModuLM enables the dynamic construction of over 50,000 distinct model configurations. In addition, we provide comprehensive results to demonstrate the effectiveness of ModuLM in supporting LLM-based MRL tasks.
- [112] arXiv:2506.00895 [pdf, html, other]
-
Title: State-Covering Trajectory Stitching for Diffusion PlannersSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Diffusion-based generative models are emerging as powerful tools for long-horizon planning in reinforcement learning (RL), particularly with offline datasets. However, their performance is fundamentally limited by the quality and diversity of training data. This often restricts their generalization to tasks outside their training distribution or longer planning horizons. To overcome this challenge, we propose State-Covering Trajectory Stitching (SCoTS), a novel reward-free trajectory augmentation method that incrementally stitches together short trajectory segments, systematically generating diverse and extended trajectories. SCoTS first learns a temporal distance-preserving latent representation that captures the underlying temporal structure of the environment, then iteratively stitches trajectory segments guided by directional exploration and novelty to effectively cover and expand this latent space. We demonstrate that SCoTS significantly improves the performance and generalization capabilities of diffusion planners on offline goal-conditioned benchmarks requiring stitching and long-horizon reasoning. Furthermore, augmented trajectories generated by SCoTS significantly improve the performance of widely used offline goal-conditioned RL algorithms across diverse environments.
- [113] arXiv:2506.00910 [pdf, html, other]
-
Title: PCoreSet: Effective Active Learning through Knowledge Distillation from Vision-Language ModelsComments: 35 pages, 30 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Knowledge distillation (KD) is a widely used framework for training compact, task-specific models by leveraging the knowledge of teacher models. However, its application to active learning (AL), which aims to minimize annotation costs through iterative sample selection, remains underexplored. This gap stems from the fact that KD typically assumes access to sufficient labeled data, whereas AL operates in data-scarce scenarios where task-specific teacher models are often unavailable. In this paper, we introduce ActiveKD, a framework that integrates AL with KD by leveraging the zero- and few-shot capabilities of large vision-language models (VLMs). A key aspect of ActiveKD is the structured prediction bias of VLMs--i.e., their predictions form clusters in the probability space. We regard this structure as an inductive bias of the teacher model, capturing generalizable output patterns beneficial to student learning. To exploit this bias, we propose Probabilistic CoreSet (PCoreSet), a selection strategy that maximizes coverage in the probability space rather than the feature space. PCoreSet strategically selects categorically diverse unlabeled samples, facilitating more efficient transfer of teacher knowledge under limited annotation budgets. Evaluations on 11 datasets show that PCoreSet consistently outperforms existing selection methods within the ActiveKD framework, advancing research at the intersection of AL and KD.
- [114] arXiv:2506.00917 [pdf, other]
-
Title: Q-learning with Posterior SamplingComments: 39 PagesSubjects: Machine Learning (cs.LG)
Bayesian posterior sampling techniques have demonstrated superior empirical performance in many exploration-exploitation settings. However, their theoretical analysis remains a challenge, especially in complex settings like reinforcement learning. In this paper, we introduce Q-Learning with Posterior Sampling (PSQL), a simple Q-learning-based algorithm that uses Gaussian posteriors on Q-values for exploration, akin to the popular Thompson Sampling algorithm in the multi-armed bandit setting. We show that in the tabular episodic MDP setting, PSQL achieves a regret bound of $\tilde O(H^2\sqrt{SAT})$, closely matching the known lower bound of $\Omega(H\sqrt{SAT})$. Here, S, A denote the number of states and actions in the underlying Markov Decision Process (MDP), and $T=KH$ with $K$ being the number of episodes and $H$ being the planning horizon. Our work provides several new technical insights into the core challenges in combining posterior sampling with dynamic programming and TD-learning-based RL algorithms, along with novel ideas for resolving those difficulties. We hope this will form a starting point for analyzing this efficient and important algorithmic technique in even more complex RL settings.
- [115] arXiv:2506.00918 [pdf, html, other]
-
Title: Principled Input-Output-Conditioned Post-Hoc Uncertainty Estimation for Regression NetworksSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Uncertainty quantification is critical in safety-sensitive applications but is often omitted from off-the-shelf neural networks due to adverse effects on predictive performance. Retrofitting uncertainty estimates post-hoc typically requires access to model parameters or gradients, limiting feasibility in practice. We propose a theoretically grounded framework for post-hoc uncertainty estimation in regression tasks by fitting an auxiliary model to both original inputs and frozen model outputs. Drawing from principles of maximum likelihood estimation and sequential parameter fitting, we formalize an exact post-hoc optimization objective that recovers the canonical MLE of Gaussian parameters, without requiring sampling or approximation at inference. While prior work has used model outputs to estimate uncertainty, we explicitly characterize the conditions under which this is valid and demonstrate the extent to which structured outputs can support quasi-epistemic inference. We find that using diverse auxiliary data, such as augmented subsets of the original training data, significantly enhances OOD detection and metric performance. Our hypothesis that frozen model outputs contain generalizable latent information about model error and predictive uncertainty is tested and confirmed. Finally, we ensure that our method maintains proper estimation of input-dependent uncertainty without relying exclusively on base model forecasts. These findings are demonstrated in toy problems and adapted to both UCI and depth regression benchmarks. Code: this https URL.
- [116] arXiv:2506.00920 [pdf, html, other]
-
Title: Position as Probability: Self-Supervised Transformers that Think Past Their Training for Length ExtrapolationComments: Note: v1: working paper; code, additional baselines, ablations, will follow in v2Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Neural and Evolutionary Computing (cs.NE)
Deep sequence models typically degrade in accuracy when test sequences significantly exceed their training lengths, yet many critical tasks--such as algorithmic reasoning, multi-step arithmetic, and compositional generalization--require robust length extrapolation. We introduce PRISM, a Probabilistic Relative-position Implicit Superposition Model, a novel positional encoding mechanism that enables Transformers to extrapolate accurately up to 10x beyond their training length. PRISM learns continuous relative positions through a differentiable histogram-filter update, preserving position uncertainty via a probabilistic superposition rather than conventional deterministic embeddings. Empirically, PRISM achieves state-of-the-art length extrapolation, successfully generalizing to previously intractable sequence lengths across algorithmic benchmarks--including arithmetic (addition, multiplication), SCAN compositionality tasks, and complex copy variants derived from DeepMind's recent datasets. Our analysis demonstrates that PRISM's stochastic positional encoding maintains sharp and interpretable internal states, providing a theoretical basis for reliable length generalization. These results advance the goal of neural sequence models that remain algorithmically robust at lengths far exceeding their training horizon.
- [117] arXiv:2506.00932 [pdf, html, other]
-
Title: Addressing the Collaboration Dilemma in Low-Data Federated Learning via Transient SparsityQiao Xiao, Boqian Wu, Andrey Poddubnyy, Elena Mocanu, Phuong H. Nguyen, Mykola Pechenizkiy, Decebal Constantin MocanuSubjects: Machine Learning (cs.LG)
Federated learning (FL) enables collaborative model training across decentralized clients while preserving data privacy, leveraging aggregated updates to build robust global models. However, this training paradigm faces significant challenges due to data heterogeneity and limited local datasets, which often impede effective collaboration. In such scenarios, we identify the Layer-wise Inertia Phenomenon in FL, wherein the middle layers of global model undergo minimal updates after early communication rounds, ultimately limiting the effectiveness of global aggregation. We demonstrate the presence of this phenomenon across a wide range of federated settings, spanning diverse datasets and architectures. To address this issue, we propose LIPS (Layer-wise Inertia Phenomenon with Sparsity), a simple yet effective method that periodically introduces transient sparsity to stimulate meaningful updates and empower global aggregation. Experiments demonstrate that LIPS effectively mitigates layer-wise inertia, enhances aggregation effectiveness, and improves overall performance in various FL scenarios. This work not only deepens the understanding of layer-wise learning dynamics in FL but also paves the way for more effective collaboration strategies in resource-constrained environments. Our code is publicly available at: this https URL.
- [118] arXiv:2506.00936 [pdf, html, other]
-
Title: Uncertainty-Aware Metabolic Stability Prediction with Dual-View Contrastive LearningPeijin Guo, Minghui Li, Hewen Pan, Bowen Chen, Yang Wu, Zikang Guo, Leo Yu Zhang, Shengshan Hu, Shengqing HuComments: This manuscript has been accepted for publication at ECML-PKDD 2025. The final version will be published in the conference proceedingsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Quantitative Methods (q-bio.QM)
Accurate prediction of molecular metabolic stability (MS) is critical for drug research and development but remains challenging due to the complex interplay of molecular interactions. Despite recent advances in graph neural networks (GNNs) for MS prediction, current approaches face two critical limitations: (1) incomplete molecular modeling due to atom-centric message-passing mechanisms that disregard bond-level topological features, and (2) prediction frameworks that lack reliable uncertainty quantification. To address these challenges, we propose TrustworthyMS, a novel contrastive learning framework designed for uncertainty-aware metabolic stability prediction. First, a molecular graph topology remapping mechanism synchronizes atom-bond interactions through edge-induced feature propagation, capturing both localized electronic effects and global conformational constraints. Second, contrastive topology-bond alignment enforces consistency between molecular topology views and bond patterns via feature alignment, enhancing representation robustness. Third, uncertainty modeling through Beta-Binomial uncertainty quantification enables simultaneous prediction and confidence calibration under epistemic uncertainty. Through extensive experiments, our results demonstrate that TrustworthyMS outperforms current state-of-the-art methods in terms of predictive performance.
- [119] arXiv:2506.00959 [pdf, html, other]
-
Title: Hidden Representation Clustering with Multi-Task Representation Learning towards Robust Online Budget AllocationSubjects: Machine Learning (cs.LG)
Marketing optimization, commonly formulated as an online budget allocation problem, has emerged as a pivotal factor in driving user growth. Most existing research addresses this problem by following the principle of 'first predict then optimize' for each individual, which presents challenges related to large-scale counterfactual prediction and solving complexity trade-offs. Note that the practical data quality is uncontrollable, and the solving scale tends to be tens of millions. Therefore, the existing approaches make the robust budget allocation non-trivial, especially in industrial scenarios with considerable data noise. To this end, this paper proposes a novel approach that solves the problem from the cluster perspective. Specifically, we propose a multi-task representation network to learn the inherent attributes of individuals and project the original features into high-dimension hidden representations through the first two layers of the trained network. Then, we divide these hidden representations into $K$ groups through partitioning-based clustering, thus reformulating the problem as an integer stochastic programming problem under different total budgets. Finally, we distill the representation module and clustering model into a multi-category model to facilitate online deployment. Offline experiments validate the effectiveness and superiority of our approach compared to six state-of-the-art marketing optimization algorithms. Online A/B tests on the Meituan platform indicate that the approach outperforms the online algorithm by 0.53% and 0.65%, considering order volume (OV) and gross merchandise volume (GMV), respectively.
- [120] arXiv:2506.00961 [pdf, html, other]
-
Title: Enhancing Parallelism in Decentralized Stochastic Convex OptimizationComments: ICML 2025Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Decentralized learning has emerged as a powerful approach for handling large datasets across multiple machines in a communication-efficient manner. However, such methods often face scalability limitations, as increasing the number of machines beyond a certain point negatively impacts convergence rates. In this work, we propose Decentralized Anytime SGD, a novel decentralized learning algorithm that significantly extends the critical parallelism threshold, enabling the effective use of more machines without compromising performance. Within the stochastic convex optimization (SCO) framework, we establish a theoretical upper bound on parallelism that surpasses the current state-of-the-art, allowing larger networks to achieve favorable statistical guarantees and closing the gap with centralized learning in highly connected topologies.
- [121] arXiv:2506.00962 [pdf, other]
-
Title: Reinforcement Learning with Random Time HorizonsSubjects: Machine Learning (cs.LG); Optimization and Control (math.OC); Probability (math.PR); Machine Learning (stat.ML)
We extend the standard reinforcement learning framework to random time horizons. While the classical setting typically assumes finite and deterministic or infinite runtimes of trajectories, we argue that multiple real-world applications naturally exhibit random (potentially trajectory-dependent) stopping times. Since those stopping times typically depend on the policy, their randomness has an effect on policy gradient formulas, which we (mostly for the first time) derive rigorously in this work both for stochastic and deterministic policies. We present two complementary perspectives, trajectory or state-space based, and establish connections to optimal control theory. Our numerical experiments demonstrate that using the proposed formulas can significantly improve optimization convergence compared to traditional approaches.
- [122] arXiv:2506.00967 [pdf, html, other]
-
Title: Pilot Contamination-Aware Graph Attention Network for Power Control in CFmMIMOSubjects: Machine Learning (cs.LG)
Optimization-based power control algorithms are predominantly iterative with high computational complexity, making them impractical for real-time applications in cell-free massive multiple-input multiple-output (CFmMIMO) systems. Learning-based methods have emerged as a promising alternative, and among them, graph neural networks (GNNs) have demonstrated their excellent performance in solving power control problems. However, all existing GNN-based approaches assume ideal orthogonality among pilot sequences for user equipments (UEs), which is unrealistic given that the number of UEs exceeds the available orthogonal pilot sequences in CFmMIMO schemes. Moreover, most learning-based methods assume a fixed number of UEs, whereas the number of active UEs varies over time in practice. Additionally, supervised training necessitates costly computational resources for computing the target power control solutions for a large volume of training samples. To address these issues, we propose a graph attention network for downlink power control in CFmMIMO systems that operates in a self-supervised manner while effectively handling pilot contamination and adapting to a dynamic number of UEs. Experimental results show its effectiveness, even in comparison to the optimal accelerated projected gradient method as a baseline.
- [123] arXiv:2506.00969 [pdf, html, other]
-
Title: Data Heterogeneity Modeling for Trustworthy Machine LearningComments: Survey paper for tutorial "Data Heterogeneity Modeling for Trustworthy Machine Learning" in KDD'25Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Data heterogeneity plays a pivotal role in determining the performance of machine learning (ML) systems. Traditional algorithms, which are typically designed to optimize average performance, often overlook the intrinsic diversity within datasets. This oversight can lead to a myriad of issues, including unreliable decision-making, inadequate generalization across different domains, unfair outcomes, and false scientific inferences. Hence, a nuanced approach to modeling data heterogeneity is essential for the development of dependable, data-driven systems. In this survey paper, we present a thorough exploration of heterogeneity-aware machine learning, a paradigm that systematically integrates considerations of data heterogeneity throughout the entire ML pipeline -- from data collection and model training to model evaluation and deployment. By applying this approach to a variety of critical fields, including healthcare, agriculture, finance, and recommendation systems, we demonstrate the substantial benefits and potential of heterogeneity-aware ML. These applications underscore how a deeper understanding of data diversity can enhance model robustness, fairness, and reliability and help model diagnosis and improvements. Moreover, we delve into future directions and provide research opportunities for the whole data mining community, aiming to promote the development of heterogeneity-aware ML.
- [124] arXiv:2506.00976 [pdf, html, other]
-
Title: Quantization-based Bounds on the Wasserstein MetricComments: 23 pages, 8 figures, 7 tablesSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
The Wasserstein metric has become increasingly important in many machine learning applications such as generative modeling, image retrieval and domain adaptation. Despite its appeal, it is often too costly to compute. This has motivated approximation methods like entropy-regularized optimal transport, downsampling, and subsampling, which trade accuracy for computational efficiency. In this paper, we consider the challenge of computing efficient approximations to the Wasserstein metric that also serve as strict upper or lower bounds. Focusing on discrete measures on regular grids, our approach involves formulating and exactly solving a Kantorovich problem on a coarse grid using a quantized measure and specially designed cost matrix, followed by an upscaling and correction stage. This is done either in the primal or dual space to obtain valid upper and lower bounds on the Wasserstein metric of the full-resolution inputs. We evaluate our methods on the DOTmark optimal transport images benchmark, demonstrating a 10x-100x speedup compared to entropy-regularized OT while keeping the approximation error below 2%.
- [125] arXiv:2506.00998 [pdf, html, other]
-
Title: LoRA-BAM: Input Filtering for Fine-tuned LLMs via Boxed Abstraction Monitors over LoRA LayersSubjects: Machine Learning (cs.LG)
Fine-tuning large language models (LLMs) improves performance on domain-specific tasks but can lead to overfitting, making them unreliable on out-of-distribution (OoD) queries. We propose LoRA-BAM - a method that adds OoD detection monitors to the LoRA layer using boxed abstraction to filter questions beyond the model's competence. Feature vectors from the fine-tuning data are extracted via the LLM and clustered. Clusters are enclosed in boxes; a question is flagged as OoD if its feature vector falls outside all boxes. To improve interpretability and robustness, we introduce a regularization loss during fine-tuning that encourages paraphrased questions to stay close in the feature space, and the enlargement of the decision boundary is based on the feature variance within a cluster. Our method complements existing defenses by providing lightweight and interpretable OoD detection.
- [126] arXiv:2506.01000 [pdf, html, other]
-
Title: Understanding Model Reprogramming for CLIP via Decoupling Visual PromptsSubjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
Model reprogramming adapts pretrained models to downstream tasks by modifying only the input and output spaces. Visual reprogramming (VR) is one instance for vision tasks that adds a trainable noise pattern (i.e., a visual prompt) to input images to facilitate downstream classification. The existing VR approaches for CLIP train a single visual prompt using all descriptions of different downstream classes. However, the limited learning capacity may result in (1) a failure to capture diverse aspects of the descriptions (e.g., shape, color, and texture), and (2) a possible bias toward less informative attributes that do not help distinguish between classes. In this paper, we introduce a decoupling-and-reweighting framework. Our decoupled visual prompts (DVP) are optimized using descriptions grouped by explicit causes (DVP-cse) or unsupervised clusters (DVP-cls). Then, we integrate the outputs of these visual prompts with a probabilistic reweighting matrix (PRM) that measures their contributions to each downstream class. Theoretically, DVP lowers the empirical risk bound. Experimentally, DVP outperforms baselines on average across 11 downstream datasets. Notably, the DVP-PRM integration enables insights into how individual visual prompts influence classification decisions, providing a probabilistic framework for understanding reprogramming. Our code is available at this https URL.
- [127] arXiv:2506.01016 [pdf, html, other]
-
Title: Optimistic critics can empower small actorsComments: RLC 2025Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Actor-critic methods have been central to many of the recent advances in deep reinforcement learning. The most common approach is to use symmetric architectures, whereby both actor and critic have the same network topology and number of parameters. However, recent works have argued for the advantages of asymmetric setups, specifically with the use of smaller actors. We perform broad empirical investigations and analyses to better understand the implications of this and find that, in general, smaller actors result in performance degradation and overfit critics. Our analyses suggest poor data collection, due to value underestimation, as one of the main causes for this behavior, and further highlight the crucial role the critic can play in alleviating this pathology. We explore techniques to mitigate the observed value underestimation, which enables further research in asymmetric actor-critic methods.
- [128] arXiv:2506.01049 [pdf, html, other]
-
Title: Taming LLMs by Scaling Learning Rates with Gradient GroupingComments: Preprint version of "Taming LLMs with Gradient Grouping" (ACL'2025). The code will be available at this https URLSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Training large language models (LLMs) poses challenges due to their massive scale and heterogeneous architectures. While adaptive optimizers like AdamW help address gradient variations, they still struggle with efficient and effective parameter-wise learning rate estimation, resulting in training instability, slow convergence, and poor compatibility with parameter-efficient fine-tuning (PEFT) techniques. This work introduces Scaling with Gradient Grouping (SGG), an optimizer wrapper that improves adaptive learning rate estimation by dynamic grouping and group-specific scaling. SGG first groups gradient statistics in each layer into clusters and then applies cluster-specific scaling to calibrate learning rates for each parameter, thus imposing collective group-wise constraints while maintaining precise per-parameter adaptation. Experiments on diverse (M)LLM benchmarks show that SGG integrates seamlessly with existing optimizers, and offers consistent gains and faster convergence over baselines, with various model sizes. Its stability across varying batch sizes and learning rates establishes SGG as a robust choice for LLM optimization.
- [129] arXiv:2506.01052 [pdf, html, other]
-
Title: A Finite-Time Analysis of TD Learning with Linear Function Approximation without Projections nor Strong ConvexitySubjects: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
We investigate the finite-time convergence properties of Temporal Difference (TD) learning with linear function approximation, a cornerstone algorithm in reinforcement learning. While prior work has established convergence guarantees, these results typically rely on the assumption that each iterate is projected onto a bounded set or that the learning rate is set according to the unknown strong convexity constant -- conditions that are both artificial and do not match the current practice.
In this paper, we challenge the necessity of such assumptions and present a refined analysis of TD learning. We show that the simple projection-free variant converges with a rate of $\tilde{\mathcal{O}}(\frac{||\theta^*||^2_2}{\sqrt{T}})$, even in the presence of Markovian noise. Our analysis reveals a novel self-bounding property of the TD updates and exploits it to guarantee bounded iterates. - [130] arXiv:2506.01054 [pdf, html, other]
-
Title: No Soundness in the Real World: On the Challenges of the Verification of Deployed Neural NetworksComments: accepted at ICML 2025. For the implementation, see this https URLSubjects: Machine Learning (cs.LG)
The ultimate goal of verification is to guarantee the safety of deployed neural networks. Here, we claim that all the state-of-the-art verifiers we are aware of fail to reach this goal. Our key insight is that theoretical soundness (bounding the full-precision output while computing with floating point) does not imply practical soundness (bounding the floating point output in a potentially stochastic environment). We prove this observation for the approaches that are currently used to achieve provable theoretical soundness, such as interval analysis and its variants. We also argue that achieving practical soundness is significantly harder computationally. We support our claims empirically as well by evaluating several well-known verification methods. To mislead the verifiers, we create adversarial networks that detect and exploit features of the deployment environment, such as the order and precision of floating point operations. We demonstrate that all the tested verifiers are vulnerable to our new deployment-specific attacks, which proves that they are not practically sound.
- [131] arXiv:2506.01059 [pdf, html, other]
-
Title: XAI-Units: Benchmarking Explainability Methods with Unit TestsJun Rui Lee, Sadegh Emami, Michael David Hollins, Timothy C. H. Wong, Carlos Ignacio Villalobos Sánchez, Francesca Toni, Dekai Zhang, Adam DejlComments: Accepted at FAccT 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Feature attribution (FA) methods are widely used in explainable AI (XAI) to help users understand how the inputs of a machine learning model contribute to its outputs. However, different FA models often provide disagreeing importance scores for the same model. In the absence of ground truth or in-depth knowledge about the inner workings of the model, it is often difficult to meaningfully determine which of the different FA methods produce more suitable explanations in different contexts. As a step towards addressing this issue, we introduce the open-source XAI-Units benchmark, specifically designed to evaluate FA methods against diverse types of model behaviours, such as feature interactions, cancellations, and discontinuous outputs. Our benchmark provides a set of paired datasets and models with known internal mechanisms, establishing clear expectations for desirable attribution scores. Accompanied by a suite of built-in evaluation metrics, XAI-Units streamlines systematic experimentation and reveals how FA methods perform against distinct, atomic kinds of model reasoning, similar to unit tests in software engineering. Crucially, by using procedurally generated models tied to synthetic datasets, we pave the way towards an objective and reliable comparison of FA methods.
- [132] arXiv:2506.01114 [pdf, other]
-
Title: Reconsidering LLM Uncertainty Estimation Methods in the WildYavuz Bakman, Duygu Nur Yaldiz, Sungmin Kang, Tuo Zhang, Baturalp Buyukates, Salman Avestimehr, Sai Praneeth KarimireddyComments: Accepted to ACL 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Large Language Model (LLM) Uncertainty Estimation (UE) methods have become a crucial tool for detecting hallucinations in recent years. While numerous UE methods have been proposed, most existing studies evaluate them in isolated short-form QA settings using threshold-independent metrics such as AUROC or PRR. However, real-world deployment of UE methods introduces several challenges. In this work, we systematically examine four key aspects of deploying UE methods in practical settings. Specifically, we assess (1) the sensitivity of UE methods to decision threshold selection, (2) their robustness to query transformations such as typos, adversarial prompts, and prior chat history, (3) their applicability to long-form generation, and (4) strategies for handling multiple UE scores for a single query. Our evaluations on 19 UE methods reveal that most of them are highly sensitive to threshold selection when there is a distribution shift in the calibration dataset. While these methods generally exhibit robustness against previous chat history and typos, they are significantly vulnerable to adversarial prompts. Additionally, while existing UE methods can be adapted for long-form generation through various strategies, there remains considerable room for improvement. Lastly, ensembling multiple UE scores at test time provides a notable performance boost, which highlights its potential as a practical improvement strategy. Code is available at: this https URL.
- [133] arXiv:2506.01115 [pdf, html, other]
-
Title: Attention Retrieves, MLP Memorizes: Disentangling Trainable Components in the TransformerSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
The Transformer architecture is central to the success of modern Large Language Models (LLMs), in part due to its surprising ability to perform a wide range of algorithmic tasks -- including mathematical reasoning, memorization, and retrieval -- using only gradient-based training on next-token prediction. While the core component of a Transformer is the self-attention mechanism, we question how much, and which aspects, of the performance gains can be attributed to it. To this end, we compare standard Transformers to variants in which either the multi-layer perceptron (MLP) layers or the attention projectors (queries and keys) are frozen at initialization. To further isolate the contribution of attention, we introduce MixiT -- the Mixing Transformer -- a simplified, principled model in which the attention coefficients are entirely random and fixed at initialization, eliminating any input-dependent computation or learning in attention. Surprisingly, we find that MixiT matches the performance of fully trained Transformers on various algorithmic tasks, especially those involving basic arithmetic or focusing heavily on memorization. For retrieval-based tasks, we observe that having input-dependent attention coefficients is consistently beneficial, while MixiT underperforms. We attribute this failure to its inability to form specialized circuits such as induction heads -- a specific circuit known to be crucial for learning and exploiting repeating patterns in input sequences. Even more interestingly, we find that attention with frozen key and query projectors is not only able to form induction heads, but can also perform competitively on language modeling. Our results underscore the importance of architectural heterogeneity, where distinct components contribute complementary inductive biases crucial for solving different classes of tasks.
- [134] arXiv:2506.01121 [pdf, html, other]
-
Title: Neuro-Symbolic Generative Diffusion Models for Physically Grounded, Robust, and Safe GenerationComments: Published at the 2nd International Conference on Neuro-symbolic Systems (NeuS 2025)Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Despite the remarkable generative capabilities of diffusion models, their integration into safety-critical or scientifically rigorous applications remains hindered by the need to ensure compliance with stringent physical, structural, and operational constraints. To address this challenge, this paper introduces Neuro-Symbolic Diffusion (NSD), a novel framework that interleaves diffusion steps with symbolic optimization, enabling the generation of certifiably consistent samples under user-defined functional and logic constraints. This key feature is provided for both standard and discrete diffusion models, enabling, for the first time, the generation of both continuous (e.g., images and trajectories) and discrete (e.g., molecular structures and natural language) outputs that comply with constraints. This ability is demonstrated on tasks spanning three key challenges: (1) Safety, in the context of non-toxic molecular generation and collision-free trajectory optimization; (2) Data scarcity, in domains such as drug discovery and materials engineering; and (3) Out-of-domain generalization, where enforcing symbolic constraints allows adaptation beyond the training distribution.
- [135] arXiv:2506.01145 [pdf, html, other]
-
Title: Slow Feature Analysis on Markov Chains from Goal-Directed BehaviorSubjects: Machine Learning (cs.LG)
Slow Feature Analysis is a unsupervised representation learning method that extracts slowly varying features from temporal data and can be used as a basis for subsequent reinforcement learning. Often, the behavior that generates the data on which the representation is learned is assumed to be a uniform random walk. Less research has focused on using samples generated by goal-directed behavior, as commonly the case in a reinforcement learning setting, to learn a representation. In a spatial setting, goal-directed behavior typically leads to significant differences in state occupancy between states that are close to a reward location and far from a reward location.
Through the perspective of optimal slow features on ergodic Markov chains, this work investigates the effects of these differences on value-function approximation in an idealized setting. Furthermore, three correction routes, which can potentially alleviate detrimental scaling effects, are evaluated and discussed. In addition, the special case of goal-averse behavior is considered. - [136] arXiv:2506.01151 [pdf, html, other]
-
Title: Earley-Driven Dynamic Pruning for Efficient Structured DecodingComments: ICML2025 posterSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Large Language Models (LLMs) have shown remarkable capabilities, yet ensuring their outputs conform to strict structural or grammatical constraints remains challenging, which is critical in function calls and domain-specific language (DSL) generation. Constrained decoding with context-free grammar is a flexible approach to guarantee LLMs' adherence to a specific format by dynamically building a token logits mask. However, creating this mask requires checking the validity of all tokens in the LLM vocabulary at every decoding step, which often incurs significant overheads in existing constrained decoding engines. To address this challenge, we propose $\textbf{ZapFormat}$, a novel $\textbf{dynamic pruning}$ strategy based on the Earley algorithm that identifies and eliminates invalid or redundant Earley states in real-time, significantly reducing memory occupation of the Earley algorithm's states. This further enables us to use a state cache to speed up structured generations on a large number of queries. We implemented ZapFormat in a new constrained decoding engine called Formatron which also incorporates existing optimizations. Through comprehensive experiments on structured generation tasks, including JSON generation, JSON Schema validation, and semantic parsing, we demonstrate that Formatron not only $\textbf{consistently maintains}$ high-precision compliant outputs but also achieves $\textbf{significant improvements}$ in inference speed up to 2x compared to state-of-the-art implementations. More importantly, Formatron is generally applicable across various LLM architectures. We release Formatron as open source at this https URL.
- [137] arXiv:2506.01153 [pdf, html, other]
-
Title: Weight-Space Linear Recurrent Neural NetworksComments: 33 pages, 21 figures, 11 tablesSubjects: Machine Learning (cs.LG)
We introduce WARP (Weight-space Adaptive Recurrent Prediction), a simple yet powerful framework that unifies weight-space learning with linear recurrence to redefine sequence modeling. Unlike conventional recurrent neural networks (RNNs) which collapse temporal dynamics into fixed-dimensional hidden states, WARP explicitly parametrizes the hidden state as the weights of a distinct root neural network. This formulation promotes higher-resolution memory, gradient-free adaptation at test-time, and seamless integration of domain-specific physical priors. Empirical validation shows that WARP matches or surpasses state-of-the-art baselines on diverse classification tasks, spanning synthetic benchmarks to real-world datasets. Furthermore, extensive experiments across sequential image completion, dynamical system reconstruction, and multivariate time series forecasting demonstrate its expressiveness and generalization capabilities. Critically, WARP's weight trajectories offer valuable insights into the model's inner workings. Ablation studies confirm the architectural necessity of key components, solidifying weight-space linear RNNs as a transformative paradigm for adaptive machine intelligence.
- [138] arXiv:2506.01158 [pdf, html, other]
-
Title: FORT: Forward-Only Regression Training of Normalizing FlowsDanyal Rehman, Oscar Davis, Jiarui Lu, Jian Tang, Michael Bronstein, Yoshua Bengio, Alexander Tong, Avishek Joey BoseComments: PreprintSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Simulation-free training frameworks have been at the forefront of the generative modelling revolution in continuous spaces, leading to neural dynamical systems that encompass modern large-scale diffusion and flow matching models. Despite the scalability of training, the generation of high-quality samples and their corresponding likelihood under the model requires expensive numerical simulation -- inhibiting adoption in numerous scientific applications such as equilibrium sampling of molecular systems. In this paper, we revisit classical normalizing flows as one-step generative models with exact likelihoods and propose a novel, scalable training objective that does not require computing the expensive change of variable formula used in conventional maximum likelihood training. We propose Forward-Only Regression Training (FORT), a simple $\ell_2$-regression objective that maps prior samples under our flow to specifically chosen targets. We demonstrate that FORT supports a wide class of targets, such as optimal transport targets and targets from pre-trained continuous-time normalizing flows (CNF). We further demonstrate that by using CNF targets, our one-step flows allow for larger-scale training that exceeds the performance and stability of maximum likelihood training, while unlocking a broader class of architectures that were previously challenging to train. Empirically, we elucidate that our trained flows can perform equilibrium conformation sampling in Cartesian coordinates of alanine dipeptide, alanine tripeptide, and alanine tetrapeptide.
- [139] arXiv:2506.01167 [pdf, html, other]
-
Title: Accelerated Learning with Linear Temporal Logic using Differentiable SimulationSubjects: Machine Learning (cs.LG); Robotics (cs.RO)
To ensure learned controllers comply with safety and reliability requirements for reinforcement learning in real-world settings remains challenging. Traditional safety assurance approaches, such as state avoidance and constrained Markov decision processes, often inadequately capture trajectory requirements or may result in overly conservative behaviors. To address these limitations, recent studies advocate the use of formal specification languages such as linear temporal logic (LTL), enabling the derivation of correct-by-construction learning objectives from the specified requirements. However, the sparse rewards associated with LTL specifications make learning extremely difficult, whereas dense heuristic-based rewards risk compromising correctness. In this work, we propose the first method, to our knowledge, that integrates LTL with differentiable simulators, facilitating efficient gradient-based learning directly from LTL specifications by coupling with differentiable paradigms. Our approach introduces soft labeling to achieve differentiable rewards and states, effectively mitigating the sparse-reward issue intrinsic to LTL without compromising objective correctness. We validate the efficacy of our method through experiments, demonstrating significant improvements in both reward attainment and training time compared to the discrete methods.
- [140] arXiv:2506.01177 [pdf, html, other]
-
Title: Bridging Quantum and Classical Computing in Drug Design: Architecture Principles for Improved Molecule GenerationComments: 10 pages, 7 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Biomolecules (q-bio.BM)
Hybrid quantum-classical machine learning offers a path to leverage noisy intermediate-scale quantum (NISQ) devices for drug discovery, but optimal model architectures remain unclear. We systematically optimize the quantum-classical bridge architecture for generative adversarial networks (GANs) in molecular discovery using multi-objective Bayesian optimization. Our optimized model (BO-QGAN) significantly improves performance, achieving a 2.27-fold higher Drug Candidate Score (DCS) than prior quantum-hybrid benchmarks and 2.21-fold higher than the classical baseline, using over 60% fewer parameters. Key findings favor layering multiple (3-4) shallow (4-8 qubit) quantum circuits sequentially, while classical architecture shows less sensitivity above a minimum capacity. This work provides the first empirically grounded architectural guidelines for hybrid models, enabling more effective integration of current quantum computers into pharmaceutical research pipelines.
- [141] arXiv:2506.01183 [pdf, html, other]
-
Title: Doubly Robust Alignment for Large Language ModelsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
This paper studies reinforcement learning from human feedback (RLHF) for aligning large language models with human preferences. While RLHF has demonstrated promising results, many algorithms are highly sensitive to misspecifications in the underlying preference model (e.g., the Bradley-Terry model), the reference policy, or the reward function, resulting in undesirable fine-tuning. To address model misspecification, we propose a doubly robust preference optimization algorithm that remains consistent when either the preference model or the reference policy is correctly specified (without requiring both). Our proposal demonstrates superior and more robust performance than state-of-the-art algorithms, both in theory and in practice. The code is available at this https URL
- [142] arXiv:2506.01194 [pdf, html, other]
-
Title: FedRPCA: Enhancing Federated LoRA Aggregation Using Robust PCADivyansh Jhunjhunwala, Arian Raje, Madan Ravi Ganesh, Chaithanya Kumar Mummadi, Chaoqun Dong, Jiawei Zhou, Wan-Yi Lin, Gauri Joshi, Zhenzhen LiSubjects: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
LoRA has emerged as one of the most promising fine-tuning techniques, especially for federated learning (FL), since it significantly reduces communication and computation costs at resource-constrained clients. However, data heterogeneity remains a significant challenge for LoRA-based FL, and the conventional aggregation strategy based on FedAvg suffers from slow convergence and suboptimal accuracy. Motivated by recent advances in model merging, particularly Task Arithmetic, we explore the idea of aggregating client LoRA parameters using scaled averaging. We first observe that a naive application of Task Arithmetic is ineffective due to the high cosine similarity between client updates, indicating significant common knowledge in the updates across clients. To address this issue, we propose decomposing client LoRA updates via Robust Principal Component Analysis (Robust-PCA) into a common low-rank component and client-specific sparse components. Our proposed algorithm FedRPCA aggregates the low-rank components through averaging, consolidating common knowledge, and applies scaled averaging to the sparse components to amplify client-specific knowledge. We evaluate our approach across a variety of vision and language tasks and demonstrate that it achieves higher final accuracy and faster convergence compared to competing baselines.
- [143] arXiv:2506.01208 [pdf, html, other]
-
Title: Multiresolution Analysis and Statistical Thresholding on Dynamic NetworksSubjects: Machine Learning (cs.LG)
Detecting structural change in dynamic network data has wide-ranging applications. Existing approaches typically divide the data into time bins, extract network features within each bin, and then compare these features over time. This introduces an inherent tradeoff between temporal resolution and the statistical stability of the extracted features. Despite this tradeoff, reminiscent of time-frequency tradeoffs in signal processing, most methods rely on a fixed temporal resolution. Choosing an appropriate resolution parameter is typically difficult and can be especially problematic in domains like cybersecurity, where anomalous behavior may emerge at multiple time scales. We address this challenge by proposing ANIE (Adaptive Network Intensity Estimation), a multi-resolution framework designed to automatically identify the time scales at which network structure evolves, enabling the joint detection of both rapid and gradual changes. Modeling interactions as Poisson processes, our method proceeds in two steps: (1) estimating a low-dimensional subspace of node behavior, and (2) deriving a set of novel empirical affinity coefficients that quantify change in interaction intensity between latent factors and support statistical testing for structural change across time scales. We provide theoretical guarantees for subspace estimation and the asymptotic behavior of the affinity coefficients, enabling model-based change detection. Experiments on synthetic networks show that ANIE adapts to the appropriate time resolution and is able to capture sharp structural changes while remaining robust to noise. Furthermore, applications to real-world data showcase the practical benefits of ANIE's multiresolution approach to detecting structural change over fixed resolution methods.
- [144] arXiv:2506.01212 [pdf, html, other]
-
Title: Dynamic Modes as Time Representation for Spatiotemporal ForecastingSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
This paper introduces a data-driven time embedding method for modeling long-range seasonal dependencies in spatiotemporal forecasting tasks. The proposed approach employs Dynamic Mode Decomposition (DMD) to extract temporal modes directly from observed data, eliminating the need for explicit timestamps or hand-crafted time features. These temporal modes serve as time representations that can be seamlessly integrated into deep spatiotemporal forecasting models. Unlike conventional embeddings such as time-of-day indicators or sinusoidal functions, our method captures complex multi-scale periodicity through spectral analysis of spatiotemporal data. Extensive experiments on urban mobility, highway traffic, and climate datasets demonstrate that the DMD-based embedding consistently improves long-horizon forecasting accuracy, reduces residual correlation, and enhances temporal generalization. The method is lightweight, model-agnostic, and compatible with any architecture that incorporates time covariates.
- [145] arXiv:2506.01213 [pdf, html, other]
-
Title: On the Stability of Graph Convolutional Neural Networks: A Probabilistic PerspectiveSubjects: Machine Learning (cs.LG); Signal Processing (eess.SP); Machine Learning (stat.ML)
Graph convolutional neural networks (GCNNs) have emerged as powerful tools for analyzing graph-structured data, achieving remarkable success across diverse applications. However, the theoretical understanding of the stability of these models, i.e., their sensitivity to small changes in the graph structure, remains in rather limited settings, hampering the development and deployment of robust and trustworthy models in practice. To fill this gap, we study how perturbations in the graph topology affect GCNN outputs and propose a novel formulation for analyzing model stability. Unlike prior studies that focus only on worst-case perturbations, our distribution-aware formulation characterizes output perturbations across a broad range of input data. This way, our framework enables, for the first time, a probabilistic perspective on the interplay between the statistical properties of the node data and perturbations in the graph topology. We conduct extensive experiments to validate our theoretical findings and demonstrate their benefits over existing baselines, in terms of both representation stability and adversarial attacks on downstream tasks. Our results demonstrate the practical significance of the proposed formulation and highlight the importance of incorporating data distribution into stability analysis.
- [146] arXiv:2506.01225 [pdf, html, other]
-
Title: Self-Refining Training for Amortized Density Functional TheorySubjects: Machine Learning (cs.LG)
Density Functional Theory (DFT) allows for predicting all the chemical and physical properties of molecular systems from first principles by finding an approximate solution to the many-body Schrödinger equation. However, the cost of these predictions becomes infeasible when increasing the scale of the energy evaluations, e.g., when calculating the ground-state energy for simulating molecular dynamics. Recent works have demonstrated that, for substantially large datasets of molecular conformations, Deep Learning-based models can predict the outputs of the classical DFT solvers by amortizing the corresponding optimization problems. In this paper, we propose a novel method that reduces the dependency of amortized DFT solvers on large pre-collected datasets by introducing a self-refining training strategy. Namely, we propose an efficient method that simultaneously trains a deep-learning model to predict the DFT outputs and samples molecular conformations that are used as training data for the model. We derive our method as a minimization of the variational upper bound on the KL-divergence measuring the discrepancy between the generated samples and the target Boltzmann distribution defined by the ground state energy. To demonstrate the utility of the proposed scheme, we perform an extensive empirical study comparing it with the models trained on the pre-collected datasets. Finally, we open-source our implementation of the proposed algorithm, optimized with asynchronous training and sampling stages, which enables simultaneous sampling and training. Code is available at this https URL.
- [147] arXiv:2506.01230 [pdf, html, other]
-
Title: Stress-Testing ML Pipelines with Adversarial Data CorruptionSubjects: Machine Learning (cs.LG)
Structured data-quality issues, such as missing values correlated with demographics, culturally biased labels, or systemic selection biases, routinely degrade the reliability of machine-learning pipelines. Regulators now increasingly demand evidence that high-stakes systems can withstand these realistic, interdependent errors, yet current robustness evaluations typically use random or overly simplistic corruptions, leaving worst-case scenarios unexplored. We introduce SAVAGE, a causally inspired framework that (i) formally models realistic data-quality issues through dependency graphs and flexible corruption templates, and (ii) systematically discovers corruption patterns that maximally degrade a target performance metric. SAVAGE employs a bi-level optimization approach to efficiently identify vulnerable data subpopulations and fine-tune corruption severity, treating the full ML pipeline, including preprocessing and potentially non-differentiable models, as a black box. Extensive experiments across multiple datasets and ML tasks (data cleaning, fairness-aware learning, uncertainty quantification) demonstrate that even a small fraction (around 5 %) of structured corruptions identified by SAVAGE severely impacts model performance, far exceeding random or manually crafted errors, and invalidating core assumptions of existing techniques. Thus, SAVAGE provides a practical tool for rigorous pipeline stress-testing, a benchmark for evaluating robustness methods, and actionable guidance for designing more resilient data workflows.
- [148] arXiv:2506.01231 [pdf, html, other]
-
Title: Towards Efficient Few-shot Graph Neural Architecture Search via Partitioning Gradient ContributionWenhao Song, Xuan Wu, Bo Yang, You Zhou, Yubin Xiao, Yanchun Liang, Hongwei Ge, Heow Pueh Lee, Chunguo WuComments: Accepted by SIGKDD 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
To address the weight coupling problem, certain studies introduced few-shot Neural Architecture Search (NAS) methods, which partition the supernet into multiple sub-supernets. However, these methods often suffer from computational inefficiency and tend to provide suboptimal partitioning schemes. To address this problem more effectively, we analyze the weight coupling problem from a novel perspective, which primarily stems from distinct modules in succeeding layers imposing conflicting gradient directions on the preceding layer modules. Based on this perspective, we propose the Gradient Contribution (GC) method that efficiently computes the cosine similarity of gradient directions among modules by decomposing the Vector-Jacobian Product during supernet backpropagation. Subsequently, the modules with conflicting gradient directions are allocated to distinct sub-supernets while similar ones are grouped together. To assess the advantages of GC and address the limitations of existing Graph Neural Architecture Search methods, which are limited to searching a single type of Graph Neural Networks (Message Passing Neural Networks (MPNNs) or Graph Transformers (GTs)), we propose the Unified Graph Neural Architecture Search (UGAS) framework, which explores optimal combinations of MPNNs and GTs. The experimental results demonstrate that GC achieves state-of-the-art (SOTA) performance in supernet partitioning quality and time efficiency. In addition, the architectures searched by UGAS+GC outperform both the manually designed GNNs and those obtained by existing NAS methods. Finally, ablation studies further demonstrate the effectiveness of all proposed methods.
- [149] arXiv:2506.01250 [pdf, html, other]
-
Title: Neural Variance-aware Dueling Bandits with Deep Representation and Shallow ExplorationSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
In this paper, we address the contextual dueling bandit problem by proposing variance-aware algorithms that leverage neural networks to approximate nonlinear utility functions. Our approach employs a \textit{variance-aware exploration strategy}, which adaptively accounts for uncertainty in pairwise comparisons while relying only on the gradients with respect to the learnable parameters of the last layer. This design effectively balances the exploration--exploitation tradeoff under both the Upper Confidence Bound (UCB) and Thompson Sampling (TS) frameworks. As a result, under standard assumptions, we establish theoretical guarantees showing that our algorithms achieve sublinear cumulative average regret of order $\bigol\lt(d \sqrt{\sum_{t=1}^T \sigma_t^2} + \sqrt{dT}\rt),$ for sufficiently wide neural networks, where $ d $ is the contextual dimension, $ \sigma_t^2 $ the variance of comparisons at round $ t $, and $ T $ the total number of rounds. We also empirically validate that our approach offers reasonable computational efficiency and achieves sublinear regret on both synthetic tasks with nonlinear utilities and real-world tasks, outperforming existing methods.
- [150] arXiv:2506.01260 [pdf, html, other]
-
Title: Protocol Models: Scaling Decentralized Training with Communication-Efficient Model ParallelismSubjects: Machine Learning (cs.LG)
Scaling models has led to significant advancements in deep learning, but training these models in decentralized settings remains challenging due to communication bottlenecks. While existing compression techniques are effective in data-parallel, they do not extend to model parallelism. Unlike data-parallel training, where weight gradients are exchanged, model-parallel requires compressing activations and activation gradients as they propagate through layers, accumulating compression errors. We propose a novel compression algorithm that compresses both forward and backward passes, enabling up to 99% compression with no convergence degradation with negligible memory/compute overhead. By leveraging a recursive structure in transformer networks, we predefine a low-dimensional subspace to confine the activations and gradients, allowing full reconstruction in subsequent layers. Our method achieves up to 100x improvement in communication efficiency and enables training billion-parameter-scale models over low-end GPUs connected via consumer-grade internet speeds as low as 80Mbps, matching the convergence of centralized datacenter systems with 100Gbps connections with model parallel.
- [151] arXiv:2506.01261 [pdf, html, other]
-
Title: The Actor-Critic Update Order Matters for PPO in Federated Reinforcement LearningSubjects: Machine Learning (cs.LG)
In the context of Federated Reinforcement Learning (FRL), applying Proximal Policy Optimization (PPO) faces challenges related to the update order of its actor and critic due to the aggregation step occurring between successive iterations. In particular, when local actors are updated based on local critic estimations, the algorithm becomes vulnerable to data heterogeneity. As a result, the conventional update order in PPO (critic first, then actor) may cause heterogeneous gradient directions among clients, hindering convergence to a globally optimal policy. To address this issue, we propose FedRAC, which reverses the update order (actor first, then critic) to eliminate the divergence of critics from different clients. Theoretical analysis shows that the convergence bound of FedRAC is immune to data heterogeneity under mild conditions, i.e., bounded level of heterogeneity and accurate policy evaluation. Empirical results indicate that the proposed algorithm obtains higher cumulative rewards and converges more rapidly in five experiments, including three classical RL environments and a highly heterogeneous autonomous driving scenario using the SUMO traffic simulator.
- [152] arXiv:2506.01290 [pdf, other]
-
Title: TSRating: Rating Quality of Diverse Time Series Data by Meta-learning from LLM JudgmentSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
High-quality time series (TS) data are essential for ensuring TS model performance, rendering research on rating TS data quality indispensable. Existing methods have shown promising rating accuracy within individual domains, primarily by extending data quality rating techniques such as influence functions and Shapley values to account for temporal characteristics. However, they neglect the fact that real-world TS data can span vastly different domains and exhibit distinct properties, hampering the accurate and efficient rating of diverse TS data. In this paper, we propose TSRating, a novel and unified framework for rating the quality of time series data crawled from diverse domains. TSRating is built on the assumption that LLMs inherit ample knowledge, acquired during their extensive pretraining, enabling them to comprehend and discern quality differences in diverse TS data. We verify this assumption by devising a series of prompts to elicit quality comparisons from LLMs for pairs of TS samples. We then fit a dedicated rating model, termed TSRater, to convert the LLMs' judgments into efficient quality predictions via TSRater's inference on future TS samples. To ensure cross-domain adaptability, we develop a meta-learning scheme to train TSRater on quality comparisons collected from nine distinct domains. To improve training efficiency, we employ signSGD for inner-loop updates, thus circumventing the demanding computation of hypergradients. Extensive experimental results on eleven benchmark datasets across three time series tasks, each using both conventional TS models and TS foundation models, demonstrate that TSRating outperforms baselines in terms of estimation accuracy, efficiency, and domain adaptability.
- [153] arXiv:2506.01302 [pdf, html, other]
-
Title: Recent Developments in GNNs for Drug DiscoverySubjects: Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
In this paper, we review recent developments and the role of Graph Neural Networks (GNNs) in computational drug discovery, including molecule generation, molecular property prediction, and drug-drug interaction prediction. By summarizing the most recent developments in this area, we underscore the capabilities of GNNs to comprehend intricate molecular patterns, while exploring both their current and prospective applications. We initiate our discussion by examining various molecular representations, followed by detailed discussions and categorization of existing GNN models based on their input types and downstream application tasks. We also collect a list of commonly used benchmark datasets for a variety of applications. We conclude the paper with brief discussions and summarize common trends in this important research area.
- [154] arXiv:2506.01303 [pdf, html, other]
-
Title: Latent Structured Hopfield Network for Semantic Association and RetrievalSubjects: Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC)
Episodic memory enables humans to recall past experiences by associating semantic elements such as objects, locations, and time into coherent event representations. While large pretrained models have shown remarkable progress in modeling semantic memory, the mechanisms for forming associative structures that support episodic memory remain underexplored. Inspired by hippocampal CA3 dynamics and its role in associative memory, we propose the Latent Structured Hopfield Network (LSHN), a biologically inspired framework that integrates continuous Hopfield attractor dynamics into an autoencoder architecture. LSHN mimics the cortical-hippocampal pathway: a semantic encoder extracts compact latent representations, a latent Hopfield network performs associative refinement through attractor convergence, and a decoder reconstructs perceptual input. Unlike traditional Hopfield networks, our model is trained end-to-end with gradient descent, achieving scalable and robust memory retrieval. Experiments on MNIST, CIFAR-10, and a simulated episodic memory task demonstrate superior performance in recalling corrupted inputs under occlusion and noise, outperforming existing associative memory models. Our work provides a computational perspective on how semantic elements can be dynamically bound into episodic memory traces through biologically grounded attractor mechanisms.
- [155] arXiv:2506.01311 [pdf, other]
-
Title: Energy Considerations for Large Pretrained Neural NetworksSubjects: Machine Learning (cs.LG)
Increasingly complex neural network architectures have achieved phenomenal performance. However, these complex models require massive computational resources that consume substantial amounts of electricity, which highlights the potential environmental impact of such models. Previous studies have demonstrated that substantial redundancies exist in large pre-trained models. However, previous work has primarily focused on compressing models while retaining comparable model performance, and the direct impact on electricity consumption appears to have received relatively little attention. By quantifying the energy usage associated with both uncompressed and compressed models, we investigate compression as a means of reducing electricity consumption. We consider nine different pre-trained models, ranging in size from 8M parameters to 138M parameters. To establish a baseline, we first train each model without compression and record the electricity usage and time required during training, along with other relevant statistics. We then apply three compression techniques: Steganographic capacity reduction, pruning, and low-rank factorization. In each of the resulting cases, we again measure the electricity usage, training time, model accuracy, and so on. We find that pruning and low-rank factorization offer no significant improvements with respect to energy usage or other related statistics, while steganographic capacity reduction provides major benefits in almost every case. We discuss the significance of these findings.
- [156] arXiv:2506.01317 [pdf, html, other]
-
Title: T-SHIRT: Token-Selective Hierarchical Data Selection for Instruction TuningComments: 17 pages, 6 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Instruction tuning is essential for Large Language Models (LLMs) to effectively follow user instructions. To improve training efficiency and reduce data redundancy, recent works use LLM-based scoring functions, e.g., Instruction-Following Difficulty (IFD), to select high-quality instruction-tuning data with scores above a threshold. While these data selection methods often lead to models that can match or even exceed the performance of models trained on the full datasets, we identify two key limitations: (i) they assess quality at the sample level, ignoring token-level informativeness; and (ii) they overlook the robustness of the scoring method, often selecting a sample due to superficial lexical features instead of its true quality. In this work, we propose Token-Selective HIeRarchical Data Selection for Instruction Tuning (T-SHIRT), a novel data selection framework that introduces a new scoring method to include only informative tokens in quality evaluation and also promotes robust and reliable samples whose neighbors also show high quality with less local inconsistencies. We demonstrate that models instruction-tuned on a curated dataset (only 5% of the original size) using T-SHIRT can outperform those trained on the entire large-scale dataset by up to 5.48 points on average across eight benchmarks. Across various LLMs and training set scales, our method consistently surpasses existing state-of-the-art data selection techniques, while also remaining both cost-effective and highly efficient. For instance, by using GPT-2 for score computation, we are able to process a dataset of 52k samples using 40 minutes on a single GPU.
- [157] arXiv:2506.01318 [pdf, html, other]
-
Title: Unlearning's Blind Spots: Over-Unlearning and Prototypical Relearning AttackComments: 10 pages, 4 figures, 2 tables, Under review at NeurIPS 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Machine unlearning (MU) aims to expunge a designated forget set from a trained model without costly retraining, yet the existing techniques overlook two critical blind spots: "over-unlearning" that deteriorates retained data near the forget set, and post-hoc "relearning" attacks that aim to resurrect the forgotten knowledge. We first derive the over-unlearning metric OU@{\epsilon}, which represents the collateral damage to the nearby region of the forget set, where the over-unlearning mainly appears. Next, we expose an unforeseen relearning threat on MU, i.e., the Prototypical Relearning Attack, which exploits the per-class prototype of the forget class with just a few samples, and easily restores the pre-unlearning performance. To counter both blind spots, we introduce Spotter, a plug-and-play objective that combines (i) a masked knowledge-distillation penalty on the nearby region of forget set to suppress OU@{\epsilon}, and (ii) an intra-class dispersion loss that scatters forget-class embeddings, neutralizing prototypical relearning attacks. On CIFAR-10, as one of validations, Spotter reduces OU@{\epsilon}by below the 0.05X of the baseline, drives forget accuracy to 0%, preserves accuracy of the retain set within 1% of difference with the original, and denies the prototype-attack by keeping the forget set accuracy within <1%, without accessing retained data. It confirms that Spotter is a practical remedy of the unlearning's blind spots.
- [158] arXiv:2506.01320 [pdf, html, other]
-
Title: $Ψ$-Sampler: Initial Particle Sampling for SMC-Based Inference-Time Reward Alignment in Score ModelsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
We introduce $\Psi$-Sampler, an SMC-based framework incorporating pCNL-based initial particle sampling for effective inference-time reward alignment with a score-based generative model. Inference-time reward alignment with score-based generative models has recently gained significant traction, following a broader paradigm shift from pre-training to post-training optimization. At the core of this trend is the application of Sequential Monte Carlo (SMC) to the denoising process. However, existing methods typically initialize particles from the Gaussian prior, which inadequately captures reward-relevant regions and results in reduced sampling efficiency. We demonstrate that initializing from the reward-aware posterior significantly improves alignment performance. To enable posterior sampling in high-dimensional latent spaces, we introduce the preconditioned Crank-Nicolson Langevin (pCNL) algorithm, which combines dimension-robust proposals with gradient-informed dynamics. This approach enables efficient and scalable posterior sampling and consistently improves performance across various reward alignment tasks, including layout-to-image generation, quantity-aware generation, and aesthetic-preference generation, as demonstrated in our experiments.
- [159] arXiv:2506.01327 [pdf, html, other]
-
Title: STSA: Federated Class-Incremental Learning via Spatial-Temporal Statistics AggregationSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Federated Class-Incremental Learning (FCIL) enables Class-Incremental Learning (CIL) from distributed data. Existing FCIL methods typically integrate old knowledge preservation into local client training. However, these methods cannot avoid spatial-temporal client drift caused by data heterogeneity and often incur significant computational and communication overhead, limiting practical deployment. To address these challenges simultaneously, we propose a novel approach, Spatial-Temporal Statistics Aggregation (STSA), which provides a unified framework to aggregate feature statistics both spatially (across clients) and temporally (across stages). The aggregated feature statistics are unaffected by data heterogeneity and can be used to update the classifier in closed form at each stage. Additionally, we introduce STSA-E, a communication-efficient variant with theoretical guarantees, achieving similar performance to STSA-E with much lower communication overhead. Extensive experiments on three widely used FCIL datasets, with varying degrees of data heterogeneity, show that our method outperforms state-of-the-art FCIL methods in terms of performance, flexibility, and both communication and computation efficiency.
- [160] arXiv:2506.01337 [pdf, html, other]
-
Title: NoiseAR: AutoRegressing Initial Noise Prior for Diffusion ModelsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Diffusion models have emerged as powerful generative frameworks, creating data samples by progressively denoising an initial random state. Traditionally, this initial state is sampled from a simple, fixed distribution like isotropic Gaussian, inherently lacking structure and a direct mechanism for external control. While recent efforts have explored ways to introduce controllability into the diffusion process, particularly at the initialization stage, they often rely on deterministic or heuristic approaches. These methods can be suboptimal, lack expressiveness, and are difficult to scale or integrate into more sophisticated optimization frameworks. In this paper, we introduce NoiseAR, a novel method for AutoRegressive Initial Noise Prior for Diffusion Models. Instead of a static, unstructured source, NoiseAR learns to generate a dynamic and controllable prior distribution for the initial noise. We formulate the generation of the initial noise prior's parameters as an autoregressive probabilistic modeling task over spatial patches or tokens. This approach enables NoiseAR to capture complex spatial dependencies and introduce learned structure into the initial state. Crucially, NoiseAR is designed to be conditional, allowing text prompts to directly influence the learned prior, thereby achieving fine-grained control over the diffusion initialization. Our experiments demonstrate that NoiseAR can generate initial noise priors that lead to improved sample quality and enhanced consistency with conditional inputs, offering a powerful, learned alternative to traditional random initialization. A key advantage of NoiseAR is its probabilistic formulation, which naturally supports seamless integration into probabilistic frameworks like Markov Decision Processes and Reinforcement Learning. Our code will be available at this https URL
- [161] arXiv:2506.01339 [pdf, html, other]
-
Title: Invariance Makes LLM Unlearning Resilient Even to Unanticipated Downstream Fine-TuningChangsheng Wang, Yihua Zhang, Jinghan Jia, Parikshit Ram, Dennis Wei, Yuguang Yao, Soumyadeep Pal, Nathalie Baracaldo, Sijia LiuComments: Accepted by ICML 2025Subjects: Machine Learning (cs.LG)
Machine unlearning offers a promising solution to privacy and safety concerns in large language models (LLMs) by selectively removing targeted knowledge while preserving utility. However, current methods are highly sensitive to downstream fine-tuning, which can quickly recover forgotten information-even from unrelated tasks. To address this, we introduce invariance into unlearning for the first time, inspired by invariant risk minimization (IRM). Building on this principle, we propose invariant LLM unlearning (ILU), a regularization-based framework that enhances robustness. Notably, ILU generalizes well to diverse fine-tuning tasks, even when trained using a single dataset. A task vector analysis is also provided to further elucidate the rationale behind ILU's effectiveness. Extensive experiments on the WMDP and MUSE benchmark, reveal that ILU significantly outperforms state-of-the-art unlearning methods, including negative preference optimization (NPO) and representation misdirection for unlearning (RMU). Notably, ILU achieves superior unlearning robustness across diverse downstream fine-tuning scenarios (e.g., math, paraphrase detection, and sentiment analysis) while preserving the fine-tuning performance.
- [162] arXiv:2506.01348 [pdf, html, other]
-
Title: Distributionally Robust Learning in Survival AnalysisSubjects: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
We introduce an innovative approach that incorporates a Distributionally Robust Learning (DRL) approach into Cox regression to enhance the robustness and accuracy of survival predictions. By formulating a DRL framework with a Wasserstein distance-based ambiguity set, we develop a variant Cox model that is less sensitive to assumptions about the underlying data distribution and more resilient to model misspecification and data perturbations. By leveraging Wasserstein duality, we reformulate the original min-max DRL problem into a tractable regularized empirical risk minimization problem, which can be computed by exponential conic programming. We provide guarantees on the finite sample behavior of our DRL-Cox model. Moreover, through extensive simulations and real world case studies, we demonstrate that our regression model achieves superior performance in terms of prediction accuracy and robustness compared with traditional methods.
- [163] arXiv:2506.01350 [pdf, html, other]
-
Title: Variational Adaptive Noise and Dropout towards Stable Recurrent Neural NetworksComments: 6 pages, 6 figures (accepted in ICDL2025)Subjects: Machine Learning (cs.LG); Robotics (cs.RO)
This paper proposes a novel stable learning theory for recurrent neural networks (RNNs), so-called variational adaptive noise and dropout (VAND). As stabilizing factors for RNNs, noise and dropout on the internal state of RNNs have been separately confirmed in previous studies. We reinterpret the optimization problem of RNNs as variational inference, showing that noise and dropout can be derived simultaneously by transforming the explicit regularization term arising in the optimization problem into implicit regularization. Their scale and ratio can also be adjusted appropriately to optimize the main objective of RNNs, respectively. In an imitation learning scenario with a mobile manipulator, only VAND is able to imitate sequential and periodic behaviors as instructed. this https URL
- [164] arXiv:2506.01352 [pdf, html, other]
-
Title: TAH-QUANT: Effective Activation Quantization in Pipeline Parallelism over Slow NetworkSubjects: Machine Learning (cs.LG)
Decentralized training of large language models offers the opportunity to pool computational resources across geographically distributed participants but faces significant network communication bottlenecks, particularly in pipeline-parallel settings. While pipeline parallelism partitions model layers across devices to handle large-scale models, it necessitates frequent communication of intermediate activations, creating challenges when network bandwidth is limited. Existing activation compression methods, such as AQ-SGD, mitigate quantization-induced errors through error compensation but impose prohibitive memory overhead by requiring storage of previous activations. To address these issues, we introduce TAH-Quant (Tile-wise Adaptive Hadamard Quantization), a novel activation quantization framework designed specifically for pipeline parallelism. Our approach integrates fine-grained tile-wise quantization for precise control, entropy-guided token-level adaptive bit allocation for optimal bit usage, and a Hadamard-based transform with pivot element swapping to effectively suppress quantization outliers. We further provide a theoretical analysis, proving that pipeline parallel training equipped with TAH-Quant maintains a convergence rate of $\mathcal{O}(1/\sqrt{T})$, matching that of vanilla stochastic gradient descent. Extensive experiments on diverse LLM tasks demonstrate that TAH-Quant achieves aggressive activation quantization (3-4 bits) ratio, which provides up to 4.3$\times$ end-to-end speedup without compromising training convergence, matches state-of-the-art methods, incurs no extra memory overhead, and generalizes well across different training scenarios.
- [165] arXiv:2506.01356 [pdf, html, other]
-
Title: Two-Stage Learning of Stabilizing Neural Controllers via Zubov Sampling and Iterative Domain ExpansionSubjects: Machine Learning (cs.LG); Robotics (cs.RO); Systems and Control (eess.SY)
Learning-based neural network (NN) control policies have shown impressive empirical performance. However, obtaining stability guarantees and estimations of the region of attraction of these learned neural controllers is challenging due to the lack of stable and scalable training and verification algorithms. Although previous works in this area have achieved great success, much conservatism remains in their framework. In this work, we propose a novel two-stage training framework to jointly synthesize the controller and Lyapunov function for continuous-time systems. By leveraging a Zubov-inspired region of attraction characterization to directly estimate stability boundaries, we propose a novel training data sampling strategy and a domain updating mechanism that significantly reduces the conservatism in training. Moreover, unlike existing works on continuous-time systems that rely on an SMT solver to formally verify the Lyapunov condition, we extend state-of-the-art neural network verifier $\alpha,\!\beta$-CROWN with the capability of performing automatic bound propagation through the Jacobian of dynamical systems and a novel verification scheme that avoids expensive bisection. To demonstrate the effectiveness of our approach, we conduct numerical experiments by synthesizing and verifying controllers on several challenging nonlinear systems across multiple dimensions. We show that our training can yield region of attractions with volume $5 - 1.5\cdot 10^{5}$ times larger compared to the baselines, and our verification on continuous systems can be up to $40-10000$ times faster compared to the traditional SMT solver dReal. Our code is available at this https URL.
- [166] arXiv:2506.01360 [pdf, html, other]
-
Title: RDB2G-Bench: A Comprehensive Benchmark for Automatic Graph Modeling of Relational DatabasesDongwon Choi, Sunwoo Kim, Juyeon Kim, Kyungho Kim, Geon Lee, Shinhwan Kang, Myunghwan Kim, Kijung ShinComments: Code and datasets are in this https URLSubjects: Machine Learning (cs.LG)
Relational databases (RDBs) are composed of interconnected tables, where relationships between them are defined through foreign keys. Recent research on applying machine learning to RDBs has explored graph-based representations of RDBs, where rows of tables are modeled as nodes, and foreign key relationships are modeled as edges. RDB-to-graph modeling helps capture cross-table dependencies, ultimately leading to enhanced performance across diverse tasks. However, there are numerous ways to model RDBs as graphs, and performance varies significantly depending on the chosen graph model. In our analysis, applying a common heuristic rule for graph modeling leads to up to a 10% drop in performance compared to the best-performing graph model, which remains non-trivial to identify. To foster research on intelligent RDB-to-graph modeling, we introduce RDB2G-Bench, the first benchmark framework for evaluating such methods. We construct extensive datasets covering 5 real-world RDBs and 12 predictive tasks, resulting in around 50k graph-performance pairs for efficient and reproducible evaluations. Thanks to our precomputed datasets, we were able to benchmark 9 automatic RDB-to-graph modeling methods on the 12 tasks over 600x faster than on-the-fly evaluation, which requires repeated model training. Our analysis of the datasets and benchmark results reveals key structural patterns affecting graph model effectiveness, along with practical implications for effective graph modeling.
- [167] arXiv:2506.01361 [pdf, html, other]
-
Title: TimeGraph: Synthetic Benchmark Datasets for Robust Time-Series Causal DiscoveryComments: 11 pages, 4 figures, accepted at KDD 2025 (Datasets and Benchmarks Track)Subjects: Machine Learning (cs.LG); Information Retrieval (cs.IR); Machine Learning (stat.ML)
Robust causal discovery in time series datasets depends on reliable benchmark datasets with known ground-truth causal relationships. However, such datasets remain scarce, and existing synthetic alternatives often overlook critical temporal properties inherent in real-world data, including nonstationarity driven by trends and seasonality, irregular sampling intervals, and the presence of unobserved confounders. To address these challenges, we introduce TimeGraph, a comprehensive suite of synthetic time-series benchmark datasets that systematically incorporates both linear and nonlinear dependencies while modeling key temporal characteristics such as trends, seasonal effects, and heterogeneous noise patterns. Each dataset is accompanied by a fully specified causal graph featuring varying densities and diverse noise distributions and is provided in two versions: one including unobserved confounders and one without, thereby offering extensive coverage of real-world complexity while preserving methodological neutrality. We further demonstrate the utility of TimeGraph through systematic evaluations of state-of-the-art causal discovery algorithms including PCMCI+, LPCMCI, and FGES across a diverse array of configurations and metrics. Our experiments reveal significant variations in algorithmic performance under realistic temporal conditions, underscoring the need for robust synthetic benchmarks in the fair and transparent assessment of causal discovery methods. The complete TimeGraph suite, including dataset generation scripts, evaluation metrics, and recommended experimental protocols, is freely available to facilitate reproducible research and foster community-driven advancements in time-series causal discovery.
- [168] arXiv:2506.01364 [pdf, html, other]
-
Title: Unraveling Spatio-Temporal Foundation Models via the Pipeline Lens: A Comprehensive ReviewYuchen Fang, Hao Miao, Yuxuan Liang, Liwei Deng, Yue Cui, Ximu Zeng, Yuyang Xia, Yan Zhao, Torben Bach Pedersen, Christian S. Jensen, Xiaofang Zhou, Kai ZhengComments: 21 pages, 10 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Spatio-temporal deep learning models aims to utilize useful patterns in such data to support tasks like prediction. However, previous deep learning models designed for specific tasks typically require separate training for each use case, leading to increased computational and storage costs. To address this issue, spatio-temporal foundation models have emerged, offering a unified framework capable of solving multiple spatio-temporal tasks. These foundation models achieve remarkable success by learning general knowledge with spatio-temporal data or transferring the general capabilities of pre-trained language models. While previous surveys have explored spatio-temporal data and methodologies separately, they have ignored a comprehensive examination of how foundation models are designed, selected, pre-trained, and adapted. As a result, the overall pipeline for spatio-temporal foundation models remains unclear. To bridge this gap, we innovatively provide an up-to-date review of previous spatio-temporal foundation models from the pipeline perspective. The pipeline begins with an introduction to different types of spatio-temporal data, followed by details of data preprocessing and embedding techniques. The pipeline then presents a novel data property taxonomy to divide existing methods according to data sources and dependencies, providing efficient and effective model design and selection for researchers. On this basis, we further illustrate the training objectives of primitive models, as well as the adaptation techniques of transferred models. Overall, our survey provides a clear and structured pipeline to understand the connection between core elements of spatio-temporal foundation models while guiding researchers to get started quickly. Additionally, we introduce emerging opportunities such as multi-objective training in the field of spatio-temporal foundation models.
- [169] arXiv:2506.01369 [pdf, html, other]
-
Title: Incentivizing LLMs to Self-Verify Their AnswersSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Large Language Models (LLMs) have demonstrated remarkable progress in complex reasoning tasks through both post-training and test-time scaling laws. While prevalent test-time scaling approaches are often realized by using external reward models to guide the model generation process, we find only marginal gains can be acquired when scaling a model post-trained on specific reasoning tasks. We identify that the limited improvement stems from distribution discrepancies between the specific post-trained generator and the general reward model. To address this, we propose a framework that incentivizes LLMs to self-verify their own answers. By unifying answer generation and verification within a single reinforcement learning (RL) process, we train models that can effectively assess the correctness of their own solutions. The trained model can further scale its performance during inference time by verifying its generations, without the need for external verifiers. We train our self-verification models based on Qwen2.5-Math-7B and DeepSeek-R1-Distill-Qwen-1.5B, demonstrating its capabilities across varying reasoning context lengths. Experiments on multiple mathematical reasoning benchmarks show that our models can not only improve post-training performance but also enable effective test-time scaling. Our code is available at this https URL.
- [170] arXiv:2506.01374 [pdf, html, other]
-
Title: Compiler Optimization via LLM Reasoning for Efficient Model ServingSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Programming Languages (cs.PL)
While model serving has unlocked unprecedented capabilities, the high cost of serving large-scale models continues to be a significant barrier to widespread accessibility and rapid innovation. Compiler optimizations have long driven substantial performance improvements, but existing compilers struggle with neural workloads due to the exponentially large and highly interdependent space of possible transformations. Although existing stochastic search techniques can be effective, they are often sample-inefficient and fail to leverage the structural context underlying compilation decisions. We set out to investigate the research question of whether reasoning with large language models (LLMs), without any retraining, can leverage the context-aware decision space of compiler optimization to significantly improve sample efficiency. To that end, we introduce a novel compilation framework (dubbed REASONING COMPILER) that formulates optimization as a sequential, context-aware decision process, guided by a large language model and structured Monte Carlo tree search (MCTS). The LLM acts as a proposal mechanism, suggesting hardware-aware transformations that reflect the current program state and accumulated performance feedback. Monte Carlo tree search (MCTS) incorporates the LLM-generated proposals to balance exploration and exploitation, facilitating structured, context-sensitive traversal of the expansive compiler optimization space. By achieving substantial speedups with markedly fewer samples than leading neural compilers, our approach demonstrates the potential of LLM-guided reasoning to transform the landscape of compiler optimization.
- [171] arXiv:2506.01376 [pdf, html, other]
-
Title: Modeling All-Atom Glycan Structures via Hierarchical Message Passing and Multi-Scale Pre-trainingComments: Published at ICML 2025. All code and data are releasedSubjects: Machine Learning (cs.LG)
Understanding the various properties of glycans with machine learning has shown some preliminary promise. However, previous methods mainly focused on modeling the backbone structure of glycans as graphs of monosaccharides (i.e., sugar units), while they neglected the atomic structures underlying each monosaccharide, which are actually important indicators of glycan properties. We fill this blank by introducing the GlycanAA model for All-Atom-wise Glycan modeling. GlycanAA models a glycan as a heterogeneous graph with monosaccharide nodes representing its global backbone structure and atom nodes representing its local atomic-level structures. Based on such a graph, GlycanAA performs hierarchical message passing to capture from local atomic-level interactions to global monosaccharide-level interactions. To further enhance model capability, we pre-train GlycanAA on a high-quality unlabeled glycan dataset, deriving the PreGlycanAA model. We design a multi-scale mask prediction algorithm to endow the model about different levels of dependencies in a glycan. Extensive benchmark results show the superiority of GlycanAA over existing glycan encoders and verify the further improvements achieved by PreGlycanAA. We maintain all resources at this https URL
- [172] arXiv:2506.01386 [pdf, html, other]
-
Title: ThinkEval: Practical Evaluation of Knowledge Preservation and Consistency in LLM Editing with Thought-based Knowledge GraphsSubjects: Machine Learning (cs.LG)
Model editing has become an important tool for addressing privacy, bias, and misinformation in large language models (LLMs) by enabling updates to knowledge without the need for retraining from scratch. However, existing editing techniques often target isolated facts, ignoring ripple effects on related knowledge, allowing edited facts to remain deducible and compromising broader contextual integrity. For example, changing Harry Potter's school from Hogwarts to Ilvermorny requires reassigning his house from Gryffindor to a suitable alternative while preserving Gryffindor's relationship with Hogwarts. In this work, we present a new model-editing setting, deep editing, to show: (1) how editing techniques fail to handle connected facts, evaluating how original knowledge sneaks through unchanged causal links, and (2) their impact on broader contextual knowledge. We introduce ThinkEval, a framework to systematically evaluate model- editing techniques by building model-specific knowledge graphs to analyze pre- and post-edit effects on fact persistence and catastrophic forgetting. We present KnowGIC, a benchmark created with ThinkEval, consisting of sequentially linked queries to measure these effects. We evaluate five editing techniques: AlphaEdit, RECT, ROME, MEMIT, and PRUNE across multiple LLMs. We find that these techniques struggle to balance indirect fact suppression with the preservation of related knowledge. Our dataset is available at: this https URL.
- [173] arXiv:2506.01387 [pdf, html, other]
-
Title: Multi Part Deployment of Neural NetworkComments: 7 pages, 1 figuresSubjects: Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
The increasing scale of modern neural networks, exemplified by architectures from IBM (530 billion neurons) and Google (500 billion parameters), presents significant challenges in terms of computational cost and infrastructure requirements. As deep neural networks continue to grow, traditional training paradigms relying on monolithic GPU clusters become increasingly unsustainable. This paper proposes a distributed system architecture that partitions a neural network across multiple servers, each responsible for a subset of neurons. Neurons are classified as local or remote, with inter-server connections managed via a metadata-driven lookup mechanism. A Multi-Part Neural Network Execution Engine facilitates seamless execution and training across distributed partitions by dynamically resolving and invoking remote neurons using stored metadata. All servers share a unified model through a network file system (NFS), ensuring consistency during parallel updates. A Neuron Distributor module enables flexible partitioning strategies based on neuron count, percentage, identifiers, or network layers. This architecture enables cost-effective, scalable deployment of deep learning models on cloud infrastructure, reducing dependency on high-performance centralized compute resources.
- [174] arXiv:2506.01393 [pdf, html, other]
-
Title: Improved Regret Bounds for Gaussian Process Upper Confidence Bound in Bayesian OptimizationComments: 37 pagesSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
This paper addresses the Bayesian optimization problem (also referred to as the Bayesian setting of the Gaussian process bandit), where the learner seeks to minimize the regret under a function drawn from a known Gaussian process (GP). Under a Matérn kernel with a certain degree of smoothness, we show that the Gaussian process upper confidence bound (GP-UCB) algorithm achieves $\tilde{O}(\sqrt{T})$ cumulative regret with high probability. Furthermore, our analysis yields $O(\sqrt{T \ln^4 T})$ regret under a squared exponential kernel. These results fill the gap between the existing regret upper bound for GP-UCB and the best-known bound provided by Scarlett (2018). The key idea in our proof is to capture the concentration behavior of the input sequence realized by GP-UCB, enabling a more refined analysis of the GP's information gain.
- [175] arXiv:2506.01396 [pdf, html, other]
-
Title: Mitigating Disparate Impact of Differentially Private Learning through Bounded Adaptive ClippingLinzh Zhao (1), Aki Rehn (1), Mikko A. Heikkilä (1), Razane Tajeddine (2), Antti Honkela (1) ((1) Department of Computer Science, University of Helsinki, Finland, (2) Department of Electrical and Computer Engineering, American University of Beirut, Lebanon)Comments: NeurIPS 2025 under review. 22 pages, 8 figuresSubjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Machine Learning (stat.ML)
Differential privacy (DP) has become an essential framework for privacy-preserving machine learning. Existing DP learning methods, however, often have disparate impacts on model predictions, e.g., for minority groups. Gradient clipping, which is often used in DP learning, can suppress larger gradients from challenging samples. We show that this problem is amplified by adaptive clipping, which will often shrink the clipping bound to tiny values to match a well-fitting majority, while significantly reducing the accuracy for others. We propose bounded adaptive clipping, which introduces a tunable lower bound to prevent excessive gradient suppression. Our method improves the accuracy of the worst-performing class on average over 10 percentage points on skewed MNIST and Fashion MNIST compared to the unbounded adaptive clipping, and over 5 percentage points over constant clipping.
- [176] arXiv:2506.01404 [pdf, html, other]
-
Title: Quantitative Error Feedback for Quantization Noise Reduction of Filtering over GraphsComments: Journal Paper from ICASSP https://doi.org/10.1109/ICASSP49660.2025.10888821Subjects: Machine Learning (cs.LG); Multiagent Systems (cs.MA); Systems and Control (eess.SY)
This paper introduces an innovative error feedback framework designed to mitigate quantization noise in distributed graph filtering, where communications are constrained to quantized messages. It comes from error spectrum shaping techniques from state-space digital filters, and therefore establishes connections between quantized filtering processes over different domains. In contrast to existing error compensation methods, our framework quantitatively feeds back the quantization noise for exact compensation. We examine the framework under three key scenarios: (i) deterministic graph filtering, (ii) graph filtering over random graphs, and (iii) graph filtering with random node-asynchronous updates. Rigorous theoretical analysis demonstrates that the proposed framework significantly reduces the effect of quantization noise, and we provide closed-form solutions for the optimal error feedback coefficients. Moreover, this quantitative error feedback mechanism can be seamlessly integrated into communication-efficient decentralized optimization frameworks, enabling lower error floors. Numerical experiments validate the theoretical results, consistently showing that our method outperforms conventional quantization strategies in terms of both accuracy and robustness.
- [177] arXiv:2506.01405 [pdf, html, other]
-
Title: SOC-DGL: Social Interaction Behavior Inspired Dual Graph Learning Framework for Drug-Target Interaction IdentificationComments: 14 pages, 17 figures (including subfigures), 4 tables. Xiang Zhao and Ruijie Li contributed equally to this work and should be considered co-first authors. The source code and datasets are available at this https URLSubjects: Machine Learning (cs.LG)
The identification of drug-target interactions (DTI) is crucial for drug discovery and repositioning, as it reveals potential uses of existing drugs, aiding in the acceleration of the drug development process and reducing associated costs. Despite the similarity information in DTI is important, most models are limited to mining direct similarity information within homogeneous graphs, overlooking the potential yet rich similarity information in heterogeneous graphs. Inspired by real-world social interaction behaviors, we propose SOC-DGL, which comprises two specialized modules: the Affinity-Driven Graph Learning (ADGL) module and the Equilibrium-Driven Graph Learning (EDGL) module. The ADGL module adopts a comprehensive social interaction strategy, leveraging an affinity-enhanced global drug-target graph to learn both global DTI and the individual similarity information of drugs and targets. In contrast, the EDGL module employs a higher-order social interaction strategy, amplifying the influence of even-hop neighbors through an even-polynomial graph filter grounded in balance theory, enabling the indirect mining of higher-order homogeneous information. This dual approach enables SOC-DGL to effectively and comprehensively capture similarity information across diverse interaction scales within the affinity matrices and drug-target association matrices, significantly enhancing the model's generalization capability and predictive accuracy in DTI tasks. To address the issue of imbalance in drug-target interaction datasets, this paper proposes an adjustable imbalance loss function that mitigates the impact of sample imbalance by adjusting the weight of negative samples and a parameter. Extensive experiments on four benchmark datasets demonstrate significant accuracy improvements achieved by SOC-DGL, particularly in scenarios involving data imbalance and unseen drugs or targets.
- [178] arXiv:2506.01414 [pdf, html, other]
-
Title: Self-supervised Latent Space Optimization with Nebula Variational CodingJournal-ref: IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. 46, no. 3, pp. 1397-1411, March 2024Subjects: Machine Learning (cs.LG); Information Theory (cs.IT)
Deep learning approaches process data in a layer-by-layer way with intermediate (or latent) features. We aim at designing a general solution to optimize the latent manifolds to improve the performance on classification, segmentation, completion and/or reconstruction through probabilistic models. This paper proposes a variational inference model which leads to a clustered embedding. We introduce additional variables in the latent space, called \textbf{nebula anchors}, that guide the latent variables to form clusters during training. To prevent the anchors from clustering among themselves, we employ the variational constraint that enforces the latent features within an anchor to form a Gaussian distribution, resulting in a generative model we refer as Nebula Variational Coding (NVC). Since each latent feature can be labeled with the closest anchor, we also propose to apply metric learning in a self-supervised way to make the separation between clusters more explicit. As a consequence, the latent variables of our variational coder form clusters which adapt to the generated semantic of the training data, \textit{e.g.} the categorical labels of each sample. We demonstrate experimentally that it can be used within different architectures designed to solve different problems including text sequence, images, 3D point clouds and volumetric data, validating the advantage of our proposed method.
- [179] arXiv:2506.01444 [pdf, html, other]
-
Title: Variance-Based Defense Against Blended Backdoor AttacksComments: This paper has been accepted at ECML PKDD 2025Subjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
Backdoor attacks represent a subtle yet effective class of cyberattacks targeting AI models, primarily due to their stealthy nature. The model behaves normally on clean data but exhibits malicious behavior only when the attacker embeds a specific trigger into the input. This attack is performed during the training phase, where the adversary corrupts a small subset of the training data by embedding a pattern and modifying the labels to a chosen target. The objective is to make the model associate the pattern with the target label while maintaining normal performance on unaltered data. Several defense mechanisms have been proposed to sanitize training data-sets. However, these methods often rely on the availability of a clean dataset to compute statistical anomalies, which may not always be feasible in real-world scenarios where datasets can be unavailable or compromised. To address this limitation, we propose a novel defense method that trains a model on the given dataset, detects poisoned classes, and extracts the critical part of the attack trigger before identifying the poisoned instances. This approach enhances explainability by explicitly revealing the harmful part of the trigger. The effectiveness of our method is demonstrated through experimental evaluations on well-known image datasets and comparative analysis against three state-of-the-art algorithms: SCAn, ABL, and AGPD.
- [180] arXiv:2506.01450 [pdf, html, other]
-
Title: ShaTS: A Shapley-based Explainability Method for Time Series Artificial Intelligence Models applied to Anomaly Detection in Industrial Internet of ThingsManuel Franco de la Peña (1), Ángel Luis Perales Gómez (1), Lorenzo Fernández Maimó (1) ((1) Departamento de Ingeniería y Tecnología de Computadores, University of Murcia, Spain, Murcia)Comments: 22 pages;16 figures;Submitted to Elsevier (Information Fusion)Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Industrial Internet of Things environments increasingly rely on advanced Anomaly Detection and explanation techniques to rapidly detect and mitigate cyberincidents, thereby ensuring operational safety. The sequential nature of data collected from these environments has enabled improvements in Anomaly Detection using Machine Learning and Deep Learning models by processing time windows rather than treating the data as tabular. However, conventional explanation methods often neglect this temporal structure, leading to imprecise or less actionable explanations. This work presents ShaTS (Shapley values for Time Series models), which is a model-agnostic explainable Artificial Intelligence method designed to enhance the precision of Shapley value explanations for time series models. ShaTS addresses the shortcomings of traditional approaches by incorporating an a priori feature grouping strategy that preserves temporal dependencies and produces both coherent and actionable insights. Experiments conducted on the SWaT dataset demonstrate that ShaTS accurately identifies critical time instants, precisely pinpoints the sensors, actuators, and processes affected by anomalies, and outperforms SHAP in terms of both explainability and resource efficiency, fulfilling the real-time requirements of industrial environments.
- [181] arXiv:2506.01467 [pdf, html, other]
-
Title: Feature-aware Hypergraph Generation via Next-Scale PredictionSubjects: Machine Learning (cs.LG); Discrete Mathematics (cs.DM)
Hypergraphs generalize traditional graphs by allowing hyperedges to connect multiple nodes, making them well-suited for modeling complex structures with higher-order relationships, such as 3D meshes, molecular systems, and electronic circuits. While topology is central to hypergraph structure, many real-world applications also require node and hyperedge features. Existing hypergraph generation methods focus solely on topology, often overlooking feature modeling. In this work, we introduce FAHNES (feature-aware hypergraph generation via next-scale prediction), a hierarchical approach that jointly generates hypergraph topology and features. FAHNES builds a multi-scale representation through node coarsening, then learns to reconstruct finer levels via localized expansion and refinement, guided by a new node budget mechanism that controls cluster splitting. We evaluate FAHNES on synthetic hypergraphs, 3D meshes, and molecular datasets. FAHNES achieves competitive results in reconstructing topology and features, establishing a foundation for future research in featured hypergraph generative modeling.
- [182] arXiv:2506.01478 [pdf, html, other]
-
Title: MUDI: A Multimodal Biomedical Dataset for Understanding Pharmacodynamic Drug-Drug InteractionsSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL); Multimedia (cs.MM); Quantitative Methods (q-bio.QM)
Understanding the interaction between different drugs (drug-drug interaction or DDI) is critical for ensuring patient safety and optimizing therapeutic outcomes. Existing DDI datasets primarily focus on textual information, overlooking multimodal data that reflect complex drug mechanisms. In this paper, we (1) introduce MUDI, a large-scale Multimodal biomedical dataset for Understanding pharmacodynamic Drug-drug Interactions, and (2) benchmark learning methods to study it. In brief, MUDI provides a comprehensive multimodal representation of drugs by combining pharmacological text, chemical formulas, molecular structure graphs, and images across 310,532 annotated drug pairs labeled as Synergism, Antagonism, or New Effect. Crucially, to effectively evaluate machine-learning based generalization, MUDI consists of unseen drug pairs in the test set. We evaluate benchmark models using both late fusion voting and intermediate fusion strategies. All data, annotations, evaluation scripts, and baselines are released under an open research license.
- [183] arXiv:2506.01482 [pdf, html, other]
-
Title: Automatic Stage Lighting Control: Is it a Rule-Driven Process or Generative Task?Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Multimedia (cs.MM); Audio and Speech Processing (eess.AS)
Stage lighting plays an essential role in live music performances, influencing the engaging experience of both musicians and audiences. Given the high costs associated with hiring or training professional lighting engineers, Automatic Stage Lighting Control (ASLC) has gained increasing attention. However, most existing approaches only classify music into limited categories and map them to predefined light patterns, resulting in formulaic and monotonous outcomes that lack rationality. To address this issue, this paper presents an end-to-end solution that directly learns from experienced lighting engineers -- Skip-BART. To the best of our knowledge, this is the first work to conceptualize ASLC as a generative task rather than merely a classification problem. Our method modifies the BART model to take audio music as input and produce light hue and value (intensity) as output, incorporating a novel skip connection mechanism to enhance the relationship between music and light within the frame this http URL validate our method through both quantitative analysis and an human evaluation, demonstrating that Skip-BART outperforms conventional rule-based methods across all evaluation metrics and shows only a limited gap compared to real lighting this http URL, our method yields a p-value of 0.72 in a statistical comparison based on human evaluations with human lighting engineers, suggesting that the proposed approach closely matches human lighting engineering performance. To support further research, we have made our self-collected dataset, code, and trained model parameters available at this https URL .
- [184] arXiv:2506.01486 [pdf, html, other]
-
Title: Model-agnostic Mitigation Strategies of Data Imbalance for RegressionComments: 34 pages, 11 figures, to be submitted to Springer Nature Machine LearningSubjects: Machine Learning (cs.LG)
Data imbalance persists as a pervasive challenge in regression tasks, introducing bias in model performance and undermining predictive reliability. This is particularly detrimental in applications aimed at predicting rare events that fall outside the domain of the bulk of the training data. In this study, we review the current state-of-the-art regarding sampling-based methods and cost-sensitive learning. Additionally, we propose novel approaches to mitigate model bias. To better asses the importance of data, we introduce the density-distance and density-ratio relevance functions, which effectively integrate empirical frequency of data with domain-specific preferences, offering enhanced interpretability for end-users. Furthermore, we present advanced mitigation techniques (cSMOGN and crbSMOGN), which build upon and improve existing sampling methods. In a comprehensive quantitative evaluation, we benchmark state-of-the-art methods on 10 synthetic and 42 real-world datasets, using neural networks, XGBoosting trees and Random Forest models. Our analysis reveals that while most strategies improve performance on rare samples, they often degrade it on frequent ones. We demonstrate that constructing an ensemble of models -- one trained with imbalance mitigation and another without -- can significantly reduce these negative effects. The key findings underscore the superior performance of our novel crbSMOGN sampling technique with the density-ratio relevance function for neural networks, outperforming state-of-the-art methods.
- [185] arXiv:2506.01490 [pdf, html, other]
-
Title: Confidence-Aware Self-Distillation for Multimodal Sentiment Analysis with Incomplete ModalitiesSubjects: Machine Learning (cs.LG)
Multimodal sentiment analysis (MSA) aims to understand human sentiment through multimodal data. In real-world scenarios, practical factors often lead to uncertain modality missingness. Existing methods for handling modality missingness are based on data reconstruction or common subspace projections. However, these methods neglect the confidence in multimodal combinations and impose constraints on intra-class representation, hindering the capture of modality-specific information and resulting in suboptimal performance. To address these challenges, we propose a Confidence-Aware Self-Distillation (CASD) strategy that effectively incorporates multimodal probabilistic embeddings via a mixture of Student's $t$-distributions, enhancing its robustness by incorporating confidence and accommodating heavy-tailed properties. This strategy estimates joint distributions with uncertainty scores and reduces uncertainty in the student network by consistency distillation. Furthermore, we introduce a reparameterization representation module that facilitates CASD in robust multimodal learning by sampling embeddings from the joint distribution for the prediction module to calculate the task loss. As a result, the directional constraint from the loss minimization is alleviated by the sampled representation. Experimental results on three benchmark datasets demonstrate that our method achieves state-of-the-art performance.
- [186] arXiv:2506.01502 [pdf, html, other]
-
Title: Learning of Population Dynamics: Inverse Optimization Meets JKO SchemeSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Learning population dynamics involves recovering the underlying process that governs particle evolution, given evolutionary snapshots of samples at discrete time points. Recent methods frame this as an energy minimization problem in probability space and leverage the celebrated JKO scheme for efficient time discretization. In this work, we introduce $\texttt{iJKOnet}$, an approach that combines the JKO framework with inverse optimization techniques to learn population dynamics. Our method relies on a conventional $\textit{end-to-end}$ adversarial training procedure and does not require restrictive architectural choices, e.g., input-convex neural networks. We establish theoretical guarantees for our methodology and demonstrate improved performance over prior JKO-based methods.
- [187] arXiv:2506.01503 [pdf, html, other]
-
Title: Analyzing the Importance of Blank for CTC-Based Knowledge DistillationComments: Accepted for Interspeech 2025Subjects: Machine Learning (cs.LG)
With the rise of large pre-trained foundation models for automatic speech recognition new challenges appear. While the performance of these models is good, runtime and cost of inference increases. One approach to make use of their strength while retaining efficiency is to distill their knowledge to smaller models during training. In this work, we explore different CTC-based distillation variants, focusing on blank token handling. We show that common approaches like blank elimination do not always work off the shelf. We explore new blank selection patterns as a potential sweet spot between standard knowledge distillation and blank elimination mechanisms. Through the introduction of a symmetric selection method, we are able to remove the CTC loss during knowledge distillation with minimal to no performance degradation. With this, we make the training independent from target labels, potentially allowing for distillation on untranscribed audio data.
- [188] arXiv:2506.01522 [pdf, html, other]
-
Title: Beyond Diagonal Covariance: Flexible Posterior VAEs via Free-Form Injective FlowsSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Variational Autoencoders (VAEs) are powerful generative models widely used for learning interpretable latent spaces, quantifying uncertainty, and compressing data for downstream generative tasks. VAEs typically rely on diagonal Gaussian posteriors due to computational constraints. Using arguments grounded in differential geometry, we demonstrate inherent limitations in the representational capacity of diagonal covariance VAEs, as illustrated by explicit low-dimensional examples. In response, we show that a regularized variant of the recently introduced Free-form Injective Flow (FIF) can be interpreted as a VAE featuring a highly flexible, implicitly defined posterior. Crucially, this regularization yields a posterior equivalent to a full Gaussian covariance distribution, yet maintains computational costs comparable to standard diagonal covariance VAEs. Experiments on image datasets validate our approach, demonstrating that incorporating full covariance substantially improves model likelihood.
- [189] arXiv:2506.01523 [pdf, html, other]
-
Title: Alignment as Distribution Learning: Your Preference Model is Explicitly a Language ModelComments: 26 pages, 7 tablesSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Alignment via reinforcement learning from human feedback (RLHF) has become the dominant paradigm for controlling the quality of outputs from large language models (LLMs). However, when viewed as `loss + regularization,' the standard RLHF objective lacks theoretical justification and incentivizes degenerate, deterministic solutions, an issue that variants such as Direct Policy Optimization (DPO) also inherit. In this paper, we rethink alignment by framing it as \emph{distribution learning} from pairwise preference feedback by explicitly modeling how information about the target language model bleeds through the preference data. This explicit modeling leads us to propose three principled learning objectives: preference maximum likelihood estimation, preference distillation, and reverse KL minimization. We theoretically show that all three approaches enjoy strong non-asymptotic $O(1/n)$ convergence to the target language model, naturally avoiding degeneracy and reward overfitting. Finally, we empirically demonstrate that our distribution learning framework, especially preference distillation, consistently outperforms or matches the performances of RLHF and DPO across various tasks and models.
- [190] arXiv:2506.01529 [pdf, html, other]
-
Title: Learning Abstract World Models with a Group-Structured Latent SpaceComments: 20 pages, 18 figuresSubjects: Machine Learning (cs.LG)
Learning meaningful abstract models of Markov Decision Processes (MDPs) is crucial for improving generalization from limited data. In this work, we show how geometric priors can be imposed on the low-dimensional representation manifold of a learned transition model. We incorporate known symmetric structures via appropriate choices of the latent space and the associated group actions, which encode prior knowledge about invariances in the environment. In addition, our framework allows the embedding of additional unstructured information alongside these symmetries. We show experimentally that this leads to better predictions of the latent transition model than fully unstructured approaches, as well as better learning on downstream RL tasks, in environments with rotational and translational features, including in first-person views of 3D environments. Additionally, our experiments show that this leads to simpler and more disentangled representations. The full code is available on GitHub to ensure reproducibility.
- [191] arXiv:2506.01533 [pdf, html, other]
-
Title: A Diffusion-Based Method for Learning the Multi-Outcome Distribution of Medical TreatmentsComments: Accepted at KDD 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
In medicine, treatments often influence multiple, interdependent outcomes, such as primary endpoints, complications, adverse events, or other secondary endpoints. Hence, to make optimal treatment decisions, clinicians are interested in learning the distribution of multi-dimensional treatment outcomes. However, the vast majority of machine learning methods for predicting treatment effects focus on single-outcome settings, despite the fact that medical data often include multiple, interdependent outcomes. To address this limitation, we propose a novel diffusion-based method called DIME to learn the joint distribution of multiple outcomes of medical treatments. We addresses three challenges relevant in medical practice: (i)it is tailored to learn the joint interventional distribution of multiple medical outcomes, which enables reliable decision-making with uncertainty quantification rather than relying solely on point estimates; (ii)it explicitly captures the dependence structure between outcomes; (iii)it can handle outcomes of mixed type, including binary, categorical, and continuous variables. In DIME, we take into account the fundamental problem of causal inference through causal masking. For training, our method decomposes the joint distribution into a series of conditional distributions with a customized conditional masking to account for the dependence structure across outcomes. For inference, our method auto-regressively generates predictions. This allows our method to move beyond point estimates of causal quantities and thus learn the joint interventional distribution. To the best of our knowledge, DIME is the first neural method tailored to learn the joint, multi-outcome distribution of medical treatments. Across various experiments, we demonstrate that our method effectively learns the joint distribution and captures shared information among multiple outcomes.
- [192] arXiv:2506.01541 [pdf, html, other]
-
Title: Adaptive Destruction Processes for Diffusion SamplersTimofei Gritsaev, Nikita Morozov, Kirill Tamogashev, Daniil Tiapkin, Sergey Samsonov, Alexey Naumov, Dmitry Vetrov, Nikolay MalkinSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
This paper explores the challenges and benefits of a trainable destruction process in diffusion samplers -- diffusion-based generative models trained to sample an unnormalised density without access to data samples. Contrary to the majority of work that views diffusion samplers as approximations to an underlying continuous-time model, we view diffusion models as discrete-time policies trained to produce samples in very few generation steps. We propose to trade some of the elegance of the underlying theory for flexibility in the definition of the generative and destruction policies. In particular, we decouple the generation and destruction variances, enabling both transition kernels to be learned as unconstrained Gaussian densities. We show that, when the number of steps is limited, training both generation and destruction processes results in faster convergence and improved sampling quality on various benchmarks. Through a robust ablation study, we investigate the design choices necessary to facilitate stable training. Finally, we show the scalability of our approach through experiments on GAN latent space sampling for conditional image generation.
- [193] arXiv:2506.01544 [pdf, html, other]
-
Title: Temporal Variational Implicit Neural RepresentationsSubjects: Machine Learning (cs.LG)
We introduce Temporal Variational Implicit Neural Representations (TV-INRs), a probabilistic framework for modeling irregular multivariate time series that enables efficient individualized imputation and forecasting. By integrating implicit neural representations with latent variable models, TV-INRs learn distributions over time-continuous generator functions conditioned on signal-specific covariates. Unlike existing approaches that require extensive training, fine-tuning or meta-learning, our method achieves accurate individualized predictions through a single forward pass. Our experiments demonstrate that with a single TV-INRs instance, we can accurately solve diverse imputation and forecasting tasks, offering a computationally efficient and scalable solution for real-world applications. TV-INRs excel especially in low-data regimes, where it outperforms existing methods by an order of magnitude in mean squared error for imputation task.
- [194] arXiv:2506.01545 [pdf, html, other]
-
Title: Class Incremental Learning for Algorithm SelectionComments: This paper was accepted at GECCO 2025. 4 pages, 2 figuresSubjects: Machine Learning (cs.LG)
Algorithm selection is commonly used to predict the best solver from a portfolio per per-instance. In many real scenarios, instances arrive in a stream: new instances become available over time, while the number of class labels can also grow as new data distributions arrive downstream. As a result, the classification model needs to be periodically updated to reflect additional solvers without catastrophic forgetting of past data. In machine-learning (ML), this is referred to as Class Incremental Learning (CIL). While commonly addressed in ML settings, its relevance to algorithm-selection in optimisation has not been previously studied. Using a bin-packing dataset, we benchmark 8 continual learning methods with respect to their ability to withstand catastrophic forgetting. We find that rehearsal-based methods significantly outperform other CIL methods. While there is evidence of forgetting, the loss is small at around 7%. Hence, these methods appear to be a viable approach to continual learning in streaming optimisation scenarios.
- [195] arXiv:2506.01552 [pdf, other]
-
Title: To Each Metric Its Decoding: Post-Hoc Optimal Decision Rules of Probabilistic Hierarchical ClassifiersComments: Accepted at ICML 2025Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Hierarchical classification offers an approach to incorporate the concept of mistake severity by leveraging a structured, labeled hierarchy. However, decoding in such settings frequently relies on heuristic decision rules, which may not align with task-specific evaluation metrics. In this work, we propose a framework for the optimal decoding of an output probability distribution with respect to a target metric. We derive optimal decision rules for increasingly complex prediction settings, providing universal algorithms when candidates are limited to the set of nodes. In the most general case of predicting a subset of nodes, we focus on rules dedicated to the hierarchical $hF_{\beta}$ scores, tailored to hierarchical settings. To demonstrate the practical utility of our approach, we conduct extensive empirical evaluations, showcasing the superiority of our proposed optimal strategies, particularly in underdetermined scenarios. These results highlight the potential of our methods to enhance the performance and reliability of hierarchical classifiers in real-world applications. The code is available at this https URL
- [196] arXiv:2506.01562 [pdf, html, other]
-
Title: Unpacking Softmax: How Temperature Drives Representation Collapse, Compression, and GeneralizationWojciech Masarczyk, Mateusz Ostaszewski, Tin Sum Cheng, Tomasz Trzciński, Aurelien Lucchi, Razvan PascanuSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
The softmax function is a fundamental building block of deep neural networks, commonly used to define output distributions in classification tasks or attention weights in transformer architectures. Despite its widespread use and proven effectiveness, its influence on learning dynamics and learned representations remains poorly understood, limiting our ability to optimize model behavior. In this paper, we study the pivotal role of the softmax function in shaping the model's representation. We introduce the concept of rank deficit bias - a phenomenon in which softmax-based deep networks find solutions of rank much lower than the number of classes. This bias depends on the softmax function's logits norm, which is implicitly influenced by hyperparameters or directly modified by softmax temperature. Furthermore, we demonstrate how to exploit the softmax dynamics to learn compressed representations or to enhance their performance on out-of-distribution data. We validate our findings across diverse architectures and real-world datasets, highlighting the broad applicability of temperature tuning in improving model performance. Our work provides new insights into the mechanisms of softmax, enabling better control over representation learning in deep neural networks.
- [197] arXiv:2506.01568 [pdf, html, other]
-
Title: Trajectory First: A Curriculum for Discovering Diverse PoliciesSubjects: Machine Learning (cs.LG); Robotics (cs.RO)
Being able to solve a task in diverse ways makes agents more robust to task variations and less prone to local optima. In this context, constrained diversity optimization has emerged as a powerful reinforcement learning (RL) framework to train a diverse set of agents in parallel. However, existing constrained-diversity RL methods often under-explore in complex tasks such as robotic manipulation, leading to a lack in policy diversity. To improve diversity optimization in RL, we therefore propose a curriculum that first explores at the trajectory level before learning step-based policies. In our empirical evaluation, we provide novel insights into the shortcoming of skill-based diversity optimization, and demonstrate empirically that our curriculum improves the diversity of the learned skills.
- [198] arXiv:2506.01569 [pdf, html, other]
-
Title: Latent Space Topology Evolution in Multilayer PerceptronsSubjects: Machine Learning (cs.LG); Algebraic Topology (math.AT)
This paper introduces a topological framework for interpreting the internal representations of Multilayer Perceptrons (MLPs). We construct a simplicial tower, a sequence of simplicial complexes connected by simplicial maps, that captures how data topology evolves across network layers. Our approach enables bi-persistence analysis: layer persistence tracks topological features within each layer across scales, while MLP persistence reveals how these features transform through the network. We prove stability theorems for our topological descriptors and establish that linear separability in latent spaces is related to disconnected components in the nerve complexes. To make our framework practical, we develop a combinatorial algorithm for computing MLP persistence and introduce trajectory-based visualisations that track data flow through the network. Experiments on synthetic and real-world medical data demonstrate our method's ability to identify redundant layers, reveal critical topological transitions, and provide interpretable insights into how MLPs progressively organise data for classification.
- [199] arXiv:2506.01582 [pdf, html, other]
-
Title: Bayes optimal learning of attention-indexed modelsSubjects: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Information Theory (cs.IT); Machine Learning (stat.ML)
We introduce the attention-indexed model (AIM), a theoretical framework for analyzing learning in deep attention layers. Inspired by multi-index models, AIM captures how token-level outputs emerge from layered bilinear interactions over high-dimensional embeddings. Unlike prior tractable attention models, AIM allows full-width key and query matrices, aligning more closely with practical transformers. Using tools from statistical mechanics and random matrix theory, we derive closed-form predictions for Bayes-optimal generalization error and identify sharp phase transitions as a function of sample complexity, model width, and sequence length. We propose a matching approximate message passing algorithm and show that gradient descent can reach optimal performance. AIM offers a solvable playground for understanding learning in modern attention architectures.
- [200] arXiv:2506.01584 [pdf, html, other]
-
Title: VirnyFlow: A Design Space for Responsible Model DevelopmentSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
Developing machine learning (ML) models requires a deep understanding of real-world problems, which are inherently multi-objective. In this paper, we present VirnyFlow, the first design space for responsible model development, designed to assist data scientists in building ML pipelines that are tailored to the specific context of their problem. Unlike conventional AutoML frameworks, VirnyFlow enables users to define customized optimization criteria, perform comprehensive experimentation across pipeline stages, and iteratively refine models in alignment with real-world constraints. Our system integrates evaluation protocol definition, multi-objective Bayesian optimization, cost-aware multi-armed bandits, query optimization, and distributed parallelism into a unified architecture. We show that VirnyFlow significantly outperforms state-of-the-art AutoML systems in both optimization quality and scalability across five real-world benchmarks, offering a flexible, efficient, and responsible alternative to black-box automation in ML development.
- [201] arXiv:2506.01594 [pdf, html, other]
-
Title: Selecting for Less Discriminatory Algorithms: A Relational Search Framework for Navigating Fairness-Accuracy Trade-offs in PracticeComments: 40 pages, 5 figures. Introduces a horizontal LDA search framework for relationally navigating fairness-accuracy trade-offs using 2021 HMDA dataSubjects: Machine Learning (cs.LG); Computers and Society (cs.CY)
As machine learning models are increasingly embedded into society through high-stakes decision-making, selecting the right algorithm for a given task, audience, and sector presents a critical challenge, particularly in the context of fairness. Traditional assessments of model fairness have often framed fairness as an objective mathematical property, treating model selection as an optimization problem under idealized informational conditions. This overlooks model multiplicity as a consideration--that multiple models can deliver similar performance while exhibiting different fairness characteristics. Legal scholars have engaged this challenge through the concept of Less Discriminatory Algorithms (LDAs), which frames model selection as a civil rights obligation. In real-world deployment, this normative challenge is bounded by constraints on fairness experimentation, e.g., regulatory standards, institutional priorities, and resource capacity.
Against these considerations, the paper revisits Lee and Floridi (2021)'s relational fairness approach using updated 2021 Home Mortgage Disclosure Act (HMDA) data, and proposes an expansion of the scope of the LDA search process. We argue that extending the LDA search horizontally, considering fairness across model families themselves, provides a lightweight complement, or alternative, to within-model hyperparameter optimization, when operationalizing fairness in non-experimental, resource constrained settings. Fairness metrics alone offer useful, but insufficient signals to accurately evaluate candidate LDAs. Rather, by using a horizontal LDA search approach with the relational trade-off framework, we demonstrate a responsible minimum viable LDA search on real-world lending outcomes. Organizations can modify this approach to systematically compare, evaluate, and select LDAs that optimize fairness and accuracy in a sector-based contextualized manner. - [202] arXiv:2506.01596 [pdf, html, other]
-
Title: Understanding and Improving Laplacian Positional Encodings For Temporal GNNsComments: ECML-PKDD 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Temporal graph learning has applications in recommendation systems, traffic forecasting, and social network analysis. Although multiple architectures have been introduced, progress in positional encoding for temporal graphs remains limited. Extending static Laplacian eigenvector approaches to temporal graphs through the supra-Laplacian has shown promise, but also poses key challenges: high eigendecomposition costs, limited theoretical understanding, and ambiguity about when and how to apply these encodings. In this paper, we address these issues by (1) offering a theoretical framework that connects supra-Laplacian encodings to per-time-slice encodings, highlighting the benefits of leveraging additional temporal connectivity, (2) introducing novel methods to reduce the computational overhead, achieving up to 56x faster runtimes while scaling to graphs with 50,000 active nodes, and (3) conducting an extensive experimental study to identify which models, tasks, and datasets benefit most from these encodings. Our findings reveal that while positional encodings can significantly boost performance in certain scenarios, their effectiveness varies across different models.
- [203] arXiv:2506.01597 [pdf, html, other]
-
Title: Policy Newton Algorithm in Reproducing Kernel Hilbert SpaceSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Reinforcement learning (RL) policies represented in Reproducing Kernel Hilbert Spaces (RKHS) offer powerful representational capabilities. While second-order optimization methods like Newton's method demonstrate faster convergence than first-order approaches, current RKHS-based policy optimization remains constrained to first-order techniques. This limitation stems primarily from the intractability of explicitly computing and inverting the infinite-dimensional Hessian operator in RKHS. We introduce Policy Newton in RKHS, the first second-order optimization framework specifically designed for RL policies represented in RKHS. Our approach circumvents direct computation of the inverse Hessian operator by optimizing a cubic regularized auxiliary objective function. Crucially, we leverage the Representer Theorem to transform this infinite-dimensional optimization into an equivalent, computationally tractable finite-dimensional problem whose dimensionality scales with the trajectory data volume. We establish theoretical guarantees proving convergence to a local optimum with a local quadratic convergence rate. Empirical evaluations on a toy financial asset allocation problem validate these theoretical properties, while experiments on standard RL benchmarks demonstrate that Policy Newton in RKHS achieves superior convergence speed and higher episodic rewards compared to established first-order RKHS approaches and parametric second-order methods. Our work bridges a critical gap between non-parametric policy representations and second-order optimization methods in reinforcement learning.
- [204] arXiv:2506.01598 [pdf, html, other]
-
Title: PMNO: A novel physics guided multi-step neural operator predictor for partial differential equationsComments: 27 pages, 12 figuresSubjects: Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
Neural operators, which aim to approximate mappings between infinite-dimensional function spaces, have been widely applied in the simulation and prediction of physical systems. However, the limited representational capacity of network architectures, combined with their heavy reliance on large-scale data, often hinder effective training and result in poor extrapolation performance. In this paper, inspired by traditional numerical methods, we propose a novel physics guided multi-step neural operator (PMNO) architecture to address these challenges in long-horizon prediction of complex physical systems. Distinct from general operator learning methods, the PMNO framework replaces the single-step input with multi-step historical data in the forward pass and introduces an implicit time-stepping scheme based on the Backward Differentiation Formula (BDF) during backpropagation. This design not only strengthens the model's extrapolation capacity but also facilitates more efficient and stable training with fewer data samples, especially for long-term predictions. Meanwhile, a causal training strategy is employed to circumvent the need for multi-stage training and to ensure efficient end-to-end optimization. The neural operator architecture possesses resolution-invariant properties, enabling the trained model to perform fast extrapolation on arbitrary spatial resolutions. We demonstrate the superior predictive performance of PMNO predictor across a diverse range of physical systems, including 2D linear system, modeling over irregular domain, complex-valued wave dynamics, and reaction-diffusion processes. Depending on the specific problem setting, various neural operator architectures, including FNO, DeepONet, and their variants, can be seamlessly integrated into the PMNO framework.
- [205] arXiv:2506.01599 [pdf, html, other]
-
Title: Connecting Neural Models Latent Geometries with Relative Geodesic RepresentationsSubjects: Machine Learning (cs.LG)
Neural models learn representations of high-dimensional data on low-dimensional manifolds. Multiple factors, including stochasticities in the training process, model architectures, and additional inductive biases, may induce different representations, even when learning the same task on the same data. However, it has recently been shown that when a latent structure is shared between distinct latent spaces, relative distances between representations can be preserved, up to distortions. Building on this idea, we demonstrate that exploiting the differential-geometric structure of latent spaces of neural models, it is possible to capture precisely the transformations between representational spaces trained on similar data distributions. Specifically, we assume that distinct neural models parametrize approximately the same underlying manifold, and introduce a representation based on the pullback metric that captures the intrinsic structure of the latent space, while scaling efficiently to large models. We validate experimentally our method on model stitching and retrieval tasks, covering autoencoders and vision foundation discriminative models, across diverse architectures, datasets, and pretraining schemes.
- [206] arXiv:2506.01614 [pdf, other]
-
Title: Contrastive Learning for Efficient Transaction Validation in UTXO-based BlockchainsComments: 6 pages, 5 figures, 3 tablesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
This paper introduces a Machine Learning (ML) approach for scalability of UTXO-based blockchains, such as Bitcoin. Prior approaches to UTXO set sharding struggle with distributing UTXOs effectively across validators, creating substantial communication overhead due to child-parent transaction dependencies. This overhead, which arises from the need to locate parent UTXOs, significantly hampers transaction processing speeds. Our solution uses ML to optimize not only UTXO set sharding but also the routing of incoming transactions, ensuring that transactions are directed to shards containing their parent UTXOs. At the heart of our approach is a framework that combines contrastive and unsupervised learning to create an embedding space for transaction outputs. This embedding allows the model to group transaction outputs based on spending relationships, making it possible to route transactions efficiently to the correct validation microservices. Trained on historical transaction data with triplet loss and online semi-hard negative mining, the model embeds parent-child spending patterns directly into its parameters, thus eliminating the need for costly, real-time parent transaction lookups. This significantly reduces cross-shard communication overhead, boosting throughput and scalability.
- [207] arXiv:2506.01625 [pdf, html, other]
-
Title: Robust Satisficing Gaussian Process Bandits Under Adversarial AttacksSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
We address the problem of Gaussian Process (GP) optimization in the presence of unknown and potentially varying adversarial perturbations. Unlike traditional robust optimization approaches that focus on maximizing performance under worst-case scenarios, we consider a robust satisficing objective, where the goal is to consistently achieve a predefined performance threshold $\tau$, even under adversarial conditions. We propose two novel algorithms based on distinct formulations of robust satisficing, and show that they are instances of a general robust satisficing framework. Further, each algorithm offers different guarantees depending on the nature of the adversary. Specifically, we derive two regret bounds: one that is sublinear over time, assuming certain conditions on the adversary and the satisficing threshold $\tau$, and another that scales with the perturbation magnitude but requires no assumptions on the adversary. Through extensive experiments, we demonstrate that our approach outperforms the established robust optimization methods in achieving the satisficing objective, particularly when the ambiguity set of the robust optimization framework is inaccurately specified.
- [208] arXiv:2506.01631 [pdf, html, other]
-
Title: Gradient-Based Model Fingerprinting for LLM Similarity Detection and Family ClassificationSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
As Large Language Models (LLMs) become integral software components in modern applications, unauthorized model derivations through fine-tuning, merging, and redistribution have emerged as critical software engineering challenges. Unlike traditional software where clone detection and license compliance are well-established, the LLM ecosystem lacks effective mechanisms to detect model lineage and enforce licensing agreements. This gap is particularly problematic when open-source model creators, such as Meta's LLaMA, require derivative works to maintain naming conventions for attribution, yet no technical means exist to verify compliance.
To fill this gap, treating LLMs as software artifacts requiring provenance tracking, we present TensorGuard, a gradient-based fingerprinting framework for LLM similarity detection and family classification. Our approach extracts model-intrinsic behavioral signatures by analyzing gradient responses to random input perturbations across tensor layers, operating independently of training data, watermarks, or specific model formats. TensorGuard supports the widely-adopted safetensors format and constructs high-dimensional fingerprints through statistical analysis of gradient features. These fingerprints enable two complementary capabilities: direct pairwise similarity assessment between arbitrary models through distance computation, and systematic family classification of unknown models via the K-Means clustering algorithm with domain-informed centroid initialization using known base models. Experimental evaluation on 58 models comprising 8 base models and 50 derivatives across five model families (Llama, Qwen, Gemma, Phi, Mistral) demonstrates 94% classification accuracy under our centroid-initialized K-Means clustering. - [209] arXiv:2506.01639 [pdf, html, other]
-
Title: Bidirectional Soft Actor-Critic: Leveraging Forward and Reverse KL Divergence for Efficient Reinforcement LearningSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
The Soft Actor-Critic (SAC) algorithm, a state-of-the-art method in maximum entropy reinforcement learning, traditionally relies on minimizing reverse Kullback-Leibler (KL) divergence for policy updates. However, this approach leads to an intractable optimal projection policy, necessitating gradient-based approximations that can suffer from instability and poor sample efficiency. This paper investigates the alternative use of forward KL divergence within SAC. We demonstrate that for Gaussian policies, forward KL divergence yields an explicit optimal projection policy -- corresponding to the mean and variance of the target Boltzmann distribution's action marginals. Building on the distinct advantages of both KL directions, we propose Bidirectional SAC, an algorithm that first initializes the policy using the explicit forward KL projection and then refines it by optimizing the reverse KL divergence. Comprehensive experiments on continuous control benchmarks show that Bidirectional SAC significantly outperforms standard SAC and other baselines, achieving up to a $30\%$ increase in episodic rewards, alongside enhanced sample efficiency.
- [210] arXiv:2506.01656 [pdf, other]
-
Title: Mixture of Experts Provably Detect and Learn the Latent Cluster Structure in Gradient-Based LearningSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Mixture of Experts (MoE), an ensemble of specialized models equipped with a router that dynamically distributes each input to appropriate experts, has achieved successful results in the field of machine learning. However, theoretical understanding of this architecture is falling behind due to its inherent complexity. In this paper, we theoretically study the sample and runtime complexity of MoE following the stochastic gradient descent (SGD) when learning a regression task with an underlying cluster structure of single index models. On the one hand, we prove that a vanilla neural network fails in detecting such a latent organization as it can only process the problem as a whole. This is intrinsically related to the concept of information exponent which is low for each cluster, but increases when we consider the entire task. On the other hand, we show that a MoE succeeds in dividing this problem into easier subproblems by leveraging the ability of each expert to weakly recover the simpler function corresponding to an individual cluster. To the best of our knowledge, this work is among the first to explore the benefits of the MoE framework by examining its SGD dynamics in the context of nonlinear regression.
- [211] arXiv:2506.01665 [pdf, html, other]
-
Title: Provably Safe Reinforcement Learning from Analytic GradientsComments: 16 pages, 10 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Robotics (cs.RO)
Deploying autonomous robots in safety-critical applications requires safety guarantees. Provably safe reinforcement learning is an active field of research which aims to provide such guarantees using safeguards. These safeguards should be integrated during training to prevent a large sim-to-real gap. While there are several approaches for safeguarding sampling-based reinforcement learning, analytic gradient-based reinforcement learning often achieves superior performance and sample efficiency. However, there is no safeguarding approach for this learning paradigm yet. Our work addresses this gap by developing the first effective safeguard for analytic gradient-based reinforcement learning. We analyse existing, differentiable safeguards, adapt them through modified mappings and gradient formulations, and integrate them with a state-of-the-art learning algorithm and a differentiable simulation. We evaluate how different safeguards affect policy optimisation using numerical experiments on two classical control tasks. The results demonstrate safeguarded training without compromising performance.
- [212] arXiv:2506.01672 [pdf, html, other]
-
Title: Minimal Impact ControlNet: Advancing Multi-ControlNet IntegrationShikun Sun, Min Zhou, Zixuan Wang, Xubin Li, Tiezheng Ge, Zijie Ye, Xiaoyu Qin, Junliang Xing, Bo Zheng, Jia JiaComments: ICLR 2025Subjects: Machine Learning (cs.LG)
With the advancement of diffusion models, there is a growing demand for high-quality, controllable image generation, particularly through methods that utilize one or multiple control signals based on ControlNet. However, in current ControlNet training, each control is designed to influence all areas of an image, which can lead to conflicts when different control signals are expected to manage different parts of the image in practical applications. This issue is especially pronounced with edge-type control conditions, where regions lacking boundary information often represent low-frequency signals, referred to as silent control signals. When combining multiple ControlNets, these silent control signals can suppress the generation of textures in related areas, resulting in suboptimal outcomes. To address this problem, we propose Minimal Impact ControlNet. Our approach mitigates conflicts through three key strategies: constructing a balanced dataset, combining and injecting feature signals in a balanced manner, and addressing the asymmetry in the score function's Jacobian matrix induced by ControlNet. These improvements enhance the compatibility of control signals, allowing for freer and more harmonious generation in areas with silent control signals.
- [213] arXiv:2506.01722 [pdf, html, other]
-
Title: When Lower-Order Terms Dominate: Adaptive Expert Algorithms for Heavy-Tailed LossesSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
We consider the problem setting of prediction with expert advice with possibly heavy-tailed losses, i.e.\ the only assumption on the losses is an upper bound on their second moments, denoted by $\theta$. We develop adaptive algorithms that do not require any prior knowledge about the range or the second moment of the losses. Existing adaptive algorithms have what is typically considered a lower-order term in their regret guarantees. We show that this lower-order term, which is often the maximum of the losses, can actually dominate the regret bound in our setting. Specifically, we show that even with small constant $\theta$, this lower-order term can scale as $\sqrt{KT}$, where $K$ is the number of experts and $T$ is the time horizon. We propose adaptive algorithms with improved regret bounds that avoid the dependence on such a lower-order term and guarantee $\mathcal{O}(\sqrt{\theta T\log(K)})$ regret in the worst case, and $\mathcal{O}(\theta \log(KT)/\Delta_{\min})$ regret when the losses are sampled i.i.d.\ from some fixed distribution, where $\Delta_{\min}$ is the difference between the mean losses of the second best expert and the best expert. Additionally, when the loss function is the squared loss, our algorithm also guarantees improved regret bounds over prior results.
- [214] arXiv:2506.01728 [pdf, html, other]
-
Title: Principled data augmentation for learning to solve quadratic programming problemsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Linear and quadratic optimization are crucial in numerous real-world applications, from training machine learning models to integer-linear optimization. Recently, learning-to-optimize methods (L2O) for linear (LPs) or quadratic programs (QPs) using message-passing graph neural networks (MPNNs) have gained traction, promising lightweight, data-driven proxies for solving such optimization problems. For example, they replace the costly computation of strong branching scores in branch-and-bound solvers, requiring solving many such optimization problems. However, robust L2O MPNNs remain challenging in data-scarce settings, especially when addressing complex optimization problems such as QPs. This work introduces a principled approach to data augmentation tailored for QPs via MPNNs. Our method leverages theoretically justified data augmentation techniques to generate diverse yet optimality-preserving instances. Furthermore, we integrate these augmentations into a self-supervised learning framework based on contrastive learning, thereby pretraining MPNNs for enhanced performance on L2O tasks. Extensive experiments demonstrate that our approach improves generalization in supervised scenarios and facilitates effective transfer learning to related optimization problems.
- [215] arXiv:2506.01741 [pdf, html, other]
-
Title: Automated Manifold Learning for Reduced Order ModelingComments: 17 pages, 6 figuresSubjects: Machine Learning (cs.LG)
The problem of identifying geometric structure in data is a cornerstone of (unsupervised) learning. As a result, Geometric Representation Learning has been widely applied across scientific and engineering domains. In this work, we investigate the use of Geometric Representation Learning for the data-driven discovery of system dynamics from spatial-temporal data. We propose to encode similarity structure in such data in a spatial-temporal proximity graph, to which we apply a range of classical and deep learning-based manifold learning approaches to learn reduced order dynamics. We observe that while manifold learning is generally capable of recovering reduced order dynamics, the quality of the learned representations varies substantially across different algorithms and hyperparameter choices. This is indicative of high sensitivity to the inherent geometric assumptions of the respective approaches and suggests a need for careful hyperparameter tuning, which can be expensive in practise. To overcome these challenges, we propose a framework for Automated Manifold Learning, which selects a manifold learning approach and corresponding hyperparameter choices based on representative subsamples of the input graph. We demonstrate that the proposed framework leads to performance gains both in scalability and in the learned representations' accuracy in capturing local and global geometric features of the underlying system dynamics.
- [216] arXiv:2506.01777 [pdf, html, other]
-
Title: DRAUN: An Algorithm-Agnostic Data Reconstruction Attack on Federated Unlearning SystemsSubjects: Machine Learning (cs.LG)
Federated Unlearning (FU) enables clients to remove the influence of specific data from a collaboratively trained shared global model, addressing regulatory requirements such as GDPR and CCPA. However, this unlearning process introduces a new privacy risk: A malicious server may exploit unlearning updates to reconstruct the data requested for removal, a form of Data Reconstruction Attack (DRA). While DRAs for machine unlearning have been studied extensively in centralized Machine Learning-as-a-Service (MLaaS) settings, their applicability to FU remains unclear due to the decentralized, client-driven nature of FU. This work presents DRAUN, the first attack framework to reconstruct unlearned data in FU systems. DRAUN targets optimization-based unlearning methods, which are widely adopted for their efficiency. We theoretically demonstrate why existing DRAs targeting machine unlearning in MLaaS fail in FU and show how DRAUN overcomes these limitations. We validate our approach through extensive experiments on four datasets and four model architectures, evaluating its performance against five popular unlearning methods, effectively demonstrating that state-of-the-art FU methods remain vulnerable to DRAs.
- [217] arXiv:2506.01780 [pdf, html, other]
-
Title: Federated Gaussian Mixture ModelsComments: 19 pages, 6 figures. Submitted to ACMSubjects: Machine Learning (cs.LG)
This paper introduces FedGenGMM, a novel one-shot federated learning approach for Gaussian Mixture Models (GMM) tailored for unsupervised learning scenarios. In federated learning (FL), where multiple decentralized clients collaboratively train models without sharing raw data, significant challenges include statistical heterogeneity, high communication costs, and privacy concerns. FedGenGMM addresses these issues by allowing local GMM models, trained independently on client devices, to be aggregated through a single communication round. This approach leverages the generative property of GMMs, enabling the creation of a synthetic dataset on the server side to train a global model efficiently. Evaluation across diverse datasets covering image, tabular, and time series data demonstrates that FedGenGMM consistently achieves performance comparable to non-federated and iterative federated methods, even under significant data heterogeneity. Additionally, FedGenGMM significantly reduces communication overhead, maintains robust performance in anomaly detection tasks, and offers flexibility in local model complexities, making it particularly suitable for edge computing environments.
- [218] arXiv:2506.01781 [pdf, html, other]
-
Title: Enhancing Customer Service Chatbots with Context-Aware NLU through Selective Attention and Multi-task LearningJournal-ref: CODS-COMAD Dec '24: 8th International Conference on Data Science and Management of Data (12th ACM IKDD CODS and 30th COMAD), Jodhpur, India, December 2024Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Customer service chatbots are conversational systems aimed at addressing customer queries, often by directing them to automated workflows. A crucial aspect of this process is the classification of the customer's intent. Presently, most intent classification models for customer care utilise only customer query for intent prediction. This may result in low-accuracy models, which cannot handle ambiguous queries. An ambiguous query like "I didn't receive my package" could indicate a delayed order, or an order that was delivered but the customer failed to receive it. Resolution of each of these scenarios requires the execution of very different sequence of steps. Utilizing additional information, such as the customer's order delivery status, in the right manner can help identify the intent for such ambiguous queries. In this paper, we have introduced a context-aware NLU model that incorporates both, the customer query and contextual information from the customer's order status for predicting customer intent. A novel selective attention module is used to extract relevant context features. We have also proposed a multi-task learning paradigm for the effective utilization of different label types available in our training data. Our suggested method, Multi-Task Learning Contextual NLU with Selective Attention Weighted Context (MTL-CNLU-SAWC), yields a 4.8% increase in top 2 accuracy score over the baseline model which only uses user queries, and a 3.5% improvement over existing state-of-the-art models that combine query and context. We have deployed our model to production for Walmart's customer care domain. Accurate intent prediction through MTL-CNLU-SAWC helps to better direct customers to automated workflows, thereby significantly reducing escalations to human agents, leading to almost a million dollars in yearly savings for the company.
- [219] arXiv:2506.01789 [pdf, other]
-
Title: Datasheets Aren't Enough: DataRubrics for Automated Quality Metrics and AccountabilityGenta Indra Winata, David Anugraha, Emmy Liu, Alham Fikri Aji, Shou-Yi Hung, Aditya Parashar, Patrick Amadeus Irawan, Ruochen Zhang, Zheng-Xin Yong, Jan Christian Blaise Cruz, Niklas Muennighoff, Seungone Kim, Hanyang Zhao, Sudipta Kar, Kezia Erina Suryoraharjo, M. Farid Adilazuarda, En-Shiun Annie Lee, Ayu Purwarianti, Derry Tanti Wijaya, Monojit ChoudhuryComments: PreprintSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Audio and Speech Processing (eess.AS)
High-quality datasets are fundamental to training and evaluating machine learning models, yet their creation-especially with accurate human annotations-remains a significant challenge. Many dataset paper submissions lack originality, diversity, or rigorous quality control, and these shortcomings are often overlooked during peer review. Submissions also frequently omit essential details about dataset construction and properties. While existing tools such as datasheets aim to promote transparency, they are largely descriptive and do not provide standardized, measurable methods for evaluating data quality. Similarly, metadata requirements at conferences promote accountability but are inconsistently enforced. To address these limitations, this position paper advocates for the integration of systematic, rubric-based evaluation metrics into the dataset review process-particularly as submission volumes continue to grow. We also explore scalable, cost-effective methods for synthetic data generation, including dedicated tools and LLM-as-a-judge approaches, to support more efficient evaluation. As a call to action, we introduce DataRubrics, a structured framework for assessing the quality of both human- and model-generated datasets. Leveraging recent advances in LLM-based evaluation, DataRubrics offers a reproducible, scalable, and actionable solution for dataset quality assessment, enabling both authors and reviewers to uphold higher standards in data-centric research. We also release code to support reproducibility of LLM-based evaluations at this https URL.
- [220] arXiv:2506.01790 [pdf, html, other]
-
Title: $IF-GUIDE$: Influence Function-Guided Detoxification of LLMsComments: Pre-printSubjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
We study how training data contributes to the emergence of toxic behaviors in large-language models. Most prior work on reducing model toxicity adopts $reactive$ approaches, such as fine-tuning pre-trained (and potentially toxic) models to align them with human values. In contrast, we propose a $proactive$ approach$-$IF-Guide$-$which leverages influence functions to identify harmful tokens within any training data and suppress their impact during training. To this end, we first show that standard influence functions are ineffective at discovering harmful training records. We then present a novel adaptation that measures token-level attributions from training data to model toxicity, along with techniques for selecting toxic training documents and a learning objective that can be integrated into both pre-training and fine-tuning. Moreover, IF-Guide does not rely on human-preference data, which is typically required by existing alignment methods. In evaluation, we demonstrate that IF-Guide substantially reduces both explicit and implicit toxicity$-$by up to 10$\times$ compared to uncensored models, and up to 3$\times$ compared to baseline alignment methods, e.g., DPO and RAD$-$across both pre-training and fine-tuning scenarios. IF-Guide is computationally efficient: a billion-parameter model is $not$ $necessary$ for computing influence scores; a million-parameter model$-$with 7.5$\times$ fewer parameters$-$can effectively serve as a proxy for identifying harmful data.
- [221] arXiv:2506.01815 [pdf, other]
-
Title: Path Signatures for Feature Extraction. An Introduction to the Mathematics Underpinning an Efficient Machine Learning TechniqueComments: 15 pages, 11 figuresSubjects: Machine Learning (cs.LG); Probability (math.PR)
We provide an introduction to the topic of path signatures as means of feature extraction for machine learning from data streams. The article stresses the mathematical theory underlying the signature methodology, highlighting the conceptual character without plunging into the technical details of rigorous proofs. These notes are based on an introductory presentation given to students of the Research Experience for Undergraduates in Industrial Mathematics and Statistics at Worcester Polytechnic Institute in June 2024.
- [222] arXiv:2506.01826 [pdf, html, other]
-
Title: Efficient Learning of Balanced Signed Graphs via Sparse Linear ProgrammingComments: 13 pages, submitted to IEEE Transactions on Signal ProcessingSubjects: Machine Learning (cs.LG); Signal Processing (eess.SP)
Signed graphs are equipped with both positive and negative edge weights, encoding pairwise correlations as well as anti-correlations in data. A balanced signed graph is a signed graph with no cycles containing an odd number of negative edges. Laplacian of a balanced signed graph has eigenvectors that map via a simple linear transform to ones in a corresponding positive graph Laplacian, thus enabling reuse of spectral filtering tools designed for positive graphs. We propose an efficient method to learn a balanced signed graph Laplacian directly from data. Specifically, extending a previous linear programming (LP) based sparse inverse covariance estimation method called CLIME, we formulate a new LP problem for each Laplacian column $i$, where the linear constraints restrict weight signs of edges stemming from node $i$, so that nodes of same / different polarities are connected by positive / negative edges. Towards optimal model selection, we derive a suitable CLIME parameter $\rho$ based on a combination of the Hannan-Quinn information criterion and a minimum feasibility criterion. We solve the LP problem efficiently by tailoring a sparse LP method based on ADMM. We theoretically prove local solution convergence of our proposed iterative algorithm. Extensive experimental results on synthetic and real-world datasets show that our balanced graph learning method outperforms competing methods and enables reuse of spectral filters, wavelets, and graph convolutional nets (GCN) constructed for positive graphs.
- [223] arXiv:2506.01827 [pdf, html, other]
-
Title: Memory Access Characterization of Large Language Models in CPU Environment and its Potential ImpactsComments: 34 pages, 14 figuresSubjects: Machine Learning (cs.LG); Hardware Architecture (cs.AR)
As machine learning algorithms are shown to be an increasingly valuable tool, the demand for their access has grown accordingly. Oftentimes, it is infeasible to run inference with larger models without an accelerator, which may be unavailable in environments that have constraints such as energy consumption, security, or cost. To increase the availability of these models, we aim to im- prove the LLM inference speed on a CPU-only environment by modifying the cache architecture. To determine what improvements could be made, we conducted two experiments using this http URL and the QWEN model: running various cache configurations and evaluating their performance, and outputting a trace of the memory footprint. Using these experiments, we investigate the memory access patterns and performance characteristics to identify potential optimizations.
- [224] arXiv:2506.01833 [pdf, html, other]
-
Title: SPACE: Your Genomic Profile Predictor is a Powerful DNA Foundation ModelComments: Accepted to ICML 2025Subjects: Machine Learning (cs.LG); Genomics (q-bio.GN)
Inspired by the success of unsupervised pre-training paradigms, researchers have applied these approaches to DNA pre-training. However, we argue that these approaches alone yield suboptimal results because pure DNA sequences lack sufficient information, since their functions are regulated by genomic profiles like chromatin accessibility. Here, we demonstrate that supervised training for genomic profile prediction serves as a more effective alternative to pure sequence pre-training. Furthermore, considering the multi-species and multi-profile nature of genomic profile prediction, we introduce our $\textbf{S}$pecies-$\textbf{P}$rofile $\textbf{A}$daptive $\textbf{C}$ollaborative $\textbf{E}$xperts (SPACE) that leverages Mixture of Experts (MoE) to better capture the relationships between DNA sequences across different species and genomic profiles, thereby learning more effective DNA representations. Through extensive experiments across various tasks, our model achieves state-of-the-art performance, establishing that DNA models trained with supervised genomic profiles serve as powerful DNA representation learners. The code is available at this https URL.
- [225] arXiv:2506.01844 [pdf, html, other]
-
Title: SmolVLA: A Vision-Language-Action Model for Affordable and Efficient RoboticsMustafa Shukor, Dana Aubakirova, Francesco Capuano, Pepijn Kooijmans, Steven Palma, Adil Zouitine, Michel Aractingi, Caroline Pascal, Martino Russi, Andres Marafioti, Simon Alibert, Matthieu Cord, Thomas Wolf, Remi CadeneComments: 24 pages. Code and assets: this https URLSubjects: Machine Learning (cs.LG); Robotics (cs.RO)
Vision-language models (VLMs) pretrained on large-scale multimodal datasets encode rich visual and linguistic knowledge, making them a strong foundation for robotics. Rather than training robotic policies from scratch, recent approaches adapt VLMs into vision-language-action (VLA) models that enable natural language-driven perception and control. However, existing VLAs are typically massive--often with billions of parameters--leading to high training costs and limited real-world deployability. Moreover, they rely on academic and industrial datasets, overlooking the growing availability of community-collected data from affordable robotic platforms. In this work, we present SmolVLA, a small, efficient, and community-driven VLA that drastically reduces both training and inference costs, while retaining competitive performance. SmolVLA is designed to be trained on a single GPU and deployed on consumer-grade GPUs or even CPUs. To further improve responsiveness, we introduce an asynchronous inference stack decoupling perception and action prediction from action execution, allowing higher control rates with chunked action generation. Despite its compact size, SmolVLA achieves performance comparable to VLAs that are 10x larger. We evaluate SmolVLA on a range of both simulated as well as real-world robotic benchmarks and release all code, pretrained models, and training data.
- [226] arXiv:2506.01849 [pdf, html, other]
-
Title: Trojan Horse Hunt in Time Series Forecasting for Space OperationsKrzysztof Kotowski, Ramez Shendy, Jakub Nalepa, Przemysław Biecek, Piotr Wilczyński, Agata Kaczmarek, Dawid Płudowski, Artur Janicki, Evridiki NtagiouSubjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
This competition hosted on Kaggle (this https URL) is the first part of a series of follow-up competitions and hackathons related to the "Assurance for Space Domain AI Applications" project funded by the European Space Agency (this https URL). The competition idea is based on one of the real-life AI security threats identified within the project -- the adversarial poisoning of continuously fine-tuned satellite telemetry forecasting models. The task is to develop methods for finding and reconstructing triggers (trojans) in advanced models for satellite telemetry forecasting used in safety-critical space operations. Participants are provided with 1) a large public dataset of real-life multivariate satellite telemetry (without triggers), 2) a reference model trained on the clean data, 3) a set of poisoned neural hierarchical interpolation (N-HiTS) models for time series forecasting trained on the dataset with injected triggers, and 4) Jupyter notebook with the training pipeline and baseline algorithm (the latter will be published in the last month of the competition). The main task of the competition is to reconstruct a set of 45 triggers (i.e., short multivariate time series segments) injected into the training data of the corresponding set of 45 poisoned models. The exact characteristics (i.e., shape, amplitude, and duration) of these triggers must be identified by participants. The popular Neural Cleanse method is adopted as a baseline, but it is not designed for time series analysis and new approaches are necessary for the task. The impact of the competition is not limited to the space domain, but also to many other safety-critical applications of advanced time series analysis where model poisoning may lead to serious consequences.
- [227] arXiv:2506.01855 [pdf, html, other]
-
Title: Trade-offs in Data Memorization via Strong Data Processing InequalitiesComments: To appear in COLT 2025Subjects: Machine Learning (cs.LG); Information Theory (cs.IT); Machine Learning (stat.ML)
Recent research demonstrated that training large language models involves memorization of a significant fraction of training data. Such memorization can lead to privacy violations when training on sensitive user data and thus motivates the study of data memorization's role in learning. In this work, we develop a general approach for proving lower bounds on excess data memorization, that relies on a new connection between strong data processing inequalities and data memorization. We then demonstrate that several simple and natural binary classification problems exhibit a trade-off between the number of samples available to a learning algorithm, and the amount of information about the training data that a learning algorithm needs to memorize to be accurate. In particular, $\Omega(d)$ bits of information about the training data need to be memorized when $O(1)$ $d$-dimensional examples are available, which then decays as the number of examples grows at a problem-specific rate. Further, our lower bounds are generally matched (up to logarithmic factors) by simple learning algorithms. We also extend our lower bounds to more general mixture-of-clusters models. Our definitions and results build on the work of Brown et al. (2021) and address several limitations of the lower bounds in their work.
- [228] arXiv:2506.01863 [pdf, html, other]
-
Title: Unified Scaling Laws for Compressed RepresentationsAndrei Panferov, Alexandra Volkova, Ionut-Vlad Modoranu, Vage Egiazarian, Mher Safaryan, Dan AlistarhComments: PreprintSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Scaling laws have shaped recent advances in machine learning by enabling predictable scaling of model performance based on model size, computation, and data volume. Concurrently, the rise in computational cost for AI has motivated model compression techniques, notably quantization and sparsification, which have emerged to mitigate the steep computational demands associated with large-scale training and inference. This paper investigates the interplay between scaling laws and compression formats, exploring whether a unified scaling framework can accurately predict model performance when training occurs over various compressed representations, such as sparse, scalar-quantized, sparse-quantized or even vector-quantized formats. Our key contributions include validating a general scaling law formulation and showing that it is applicable both individually but also composably across compression types. Based on this, our main finding is demonstrating both theoretically and empirically that there exists a simple "capacity" metric -- based on the representation's ability to fit random Gaussian data -- which can robustly predict parameter efficiency across multiple compressed representations. On the practical side, we extend our formulation to directly compare the accuracy potential of different compressed formats, and to derive better algorithms for training over sparse-quantized formats.
- [229] arXiv:2506.01868 [pdf, html, other]
-
Title: NepTrain and NepTrainKit: Automated Active Learning and Visualization Toolkit for Neuroevolution PotentialsSubjects: Machine Learning (cs.LG); Materials Science (cond-mat.mtrl-sci)
As a machine-learned potential, the neuroevolution potential (NEP) method features exceptional computational efficiency and has been successfully applied in materials science. Constructing high-quality training datasets is crucial for developing accurate NEP models. However, the preparation and screening of NEP training datasets remain a bottleneck for broader applications due to their time-consuming, labor-intensive, and resource-intensive nature. In this work, we have developed NepTrain and NepTrainKit, which are dedicated to initializing and managing training datasets to generate high-quality training sets while automating NEP model training. NepTrain is an open-source Python package that features a bond length filtering method to effectively identify and remove non-physical structures from molecular dynamics trajectories, thereby ensuring high-quality training datasets. NepTrainKit is a graphical user interface (GUI) software designed specifically for NEP training datasets, providing functionalities for data editing, visualization, and interactive exploration. It integrates key features such as outlier identification, farthest-point sampling, non-physical structure detection, and configuration type selection. The combination of these tools enables users to process datasets more efficiently and conveniently. Using $\rm CsPbI_3$ as a case study, we demonstrate the complete workflow for training NEP models with NepTrain and further validate the models through materials property predictions. We believe this toolkit will greatly benefit researchers working with machine learning interatomic potentials.
- [230] arXiv:2506.01869 [pdf, html, other]
-
Title: Frugal Machine Learning for Energy-efficient, and Resource-aware Artificial IntelligenceSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Frugal Machine Learning (FML) refers to the practice of designing Machine Learning (ML) models that are efficient, cost-effective, and mindful of resource constraints. This field aims to achieve acceptable performance while minimizing the use of computational resources, time, energy, and data for both training and inference. FML strategies can be broadly categorized into input frugality, learning process frugality, and model frugality, each focusing on reducing resource consumption at different stages of the ML pipeline. This chapter explores recent advancements, applications, and open challenges in FML, emphasizing its importance for smart environments that incorporate edge computing and IoT devices, which often face strict limitations in bandwidth, energy, or latency. Technological enablers such as model compression, energy-efficient hardware, and data-efficient learning techniques are discussed, along with adaptive methods including parameter regularization, knowledge distillation, and dynamic architecture design that enable incremental model updates without full retraining. Furthermore, it provides a comprehensive taxonomy of frugal methods, discusses case studies across diverse domains, and identifies future research directions to drive innovation in this evolving field.
- [231] arXiv:2506.01876 [pdf, html, other]
-
Title: Learning to Explore: An In-Context Learning Approach for Pure ExplorationSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
In this work, we study the active sequential hypothesis testing problem, also known as pure exploration, where the goal is to actively control a data collection process to efficiently identify the correct hypothesis underlying a decision problem. While relevant across multiple domains, devising adaptive exploration strategies remains challenging, particularly due to difficulties in encoding appropriate inductive biases. Existing Reinforcement Learning (RL)-based methods often underperform when relevant information structures are inadequately represented, whereas more complex methods, like Best Arm Identification (BAI) techniques, may be difficult to devise and typically rely on explicit modeling assumptions. To address these limitations, we introduce In-Context Pure Exploration (ICPE), an in-context learning approach that uses Transformers to learn exploration strategies directly from experience. ICPE combines supervised learning and reinforcement learning to identify and exploit latent structure across related tasks, without requiring prior assumptions. Numerical results across diverse synthetic and semi-synthetic benchmarks highlight ICPE's capability to achieve robust performance performance in deterministic, stochastic, and structured settings. These results demonstrate ICPE's ability to match optimal instance-dependent algorithms using only deep learning techniques, making it a practical and general approach to data-efficient exploration.
- [232] arXiv:2506.01883 [pdf, html, other]
-
Title: scDataset: Scalable Data Loading for Deep Learning on Large-Scale Single-Cell OmicsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Databases (cs.DB)
Modern single-cell datasets now comprise hundreds of millions of cells, presenting significant challenges for training deep learning models that require shuffled, memory-efficient data loading. While the AnnData format is the community standard for storing single-cell datasets, existing data loading solutions for AnnData are often inadequate: some require loading all data into memory, others convert to dense formats that increase storage demands, and many are hampered by slow random disk access. We present scDataset, a PyTorch IterableDataset that operates directly on one or more AnnData files without the need for format conversion. The core innovation is a combination of block sampling and batched fetching, which together balance randomness and I/O efficiency. On the Tahoe 100M dataset, scDataset achieves up to a 48$\times$ speed-up over AnnLoader, a 27$\times$ speed-up over HuggingFace Datasets, and an 18$\times$ speed-up over BioNeMo in single-core settings. These advances democratize large-scale single-cell model training for the broader research community.
- [233] arXiv:2506.01884 [pdf, other]
-
Title: Agnostic Reinforcement Learning: Foundations and AlgorithmsComments: Ph.D. thesisSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Reinforcement Learning (RL) has demonstrated tremendous empirical success across numerous challenging domains. However, we lack a strong theoretical understanding of the statistical complexity of RL in environments with large state spaces, where function approximation is required for sample-efficient learning. This thesis addresses this gap by rigorously examining the statistical complexity of RL with function approximation from a learning theoretic perspective. Departing from a long history of prior work, we consider the weakest form of function approximation, called agnostic policy learning, in which the learner seeks to find the best policy in a given class $\Pi$, with no guarantee that $\Pi$ contains an optimal policy for the underlying task.
We systematically explore agnostic policy learning along three key axes: environment access -- how a learner collects data from the environment; coverage conditions -- intrinsic properties of the underlying MDP measuring the expansiveness of state-occupancy measures for policies in the class $\Pi$, and representational conditions -- structural assumptions on the class $\Pi$ itself. Within this comprehensive framework, we (1) design new learning algorithms with theoretical guarantees and (2) characterize fundamental performance bounds of any algorithm. Our results reveal significant statistical separations that highlight the power and limitations of agnostic policy learning. - [234] arXiv:2506.01890 [pdf, html, other]
-
Title: CogniAlign: Word-Level Multimodal Speech Alignment with Gated Cross-Attention for Alzheimer's DetectionSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Early detection of cognitive disorders such as Alzheimer's disease is critical for enabling timely clinical intervention and improving patient outcomes. In this work, we introduce CogniAlign, a multimodal architecture for Alzheimer's detection that integrates audio and textual modalities, two non-intrusive sources of information that offer complementary insights into cognitive health. Unlike prior approaches that fuse modalities at a coarse level, CogniAlign leverages a word-level temporal alignment strategy that synchronizes audio embeddings with corresponding textual tokens based on transcription timestamps. This alignment supports the development of token-level fusion techniques, enabling more precise cross-modal interactions. To fully exploit this alignment, we propose a Gated Cross-Attention Fusion mechanism, where audio features attend over textual representations, guided by the superior unimodal performance of the text modality. In addition, we incorporate prosodic cues, specifically interword pauses, by inserting pause tokens into the text and generating audio embeddings for silent intervals, further enriching both streams. We evaluate CogniAlign on the ADReSSo dataset, where it achieves an accuracy of 90.36%, outperforming existing state-of-the-art methods. A detailed ablation study confirms the advantages of our alignment strategy, attention-based fusion, and prosodic modeling.
- [235] arXiv:2506.01897 [pdf, html, other]
-
Title: MLorc: Momentum Low-rank Compression for Large Language Model AdaptationSubjects: Machine Learning (cs.LG); Information Theory (cs.IT); Optimization and Control (math.OC)
With increasing size of large language models (LLMs), full-parameter fine-tuning imposes substantial memory demands. To alleviate this, we propose a novel memory-efficient training paradigm called Momentum Low-rank compression (MLorc). By directly compressing and reconstructing momentum rather than gradients, MLorc avoids imposing a fixed-rank constraint on weight update matrices and better preserves the training dynamics of full-parameter fine-tuning, in contrast to existing low-rank approaches such as LoRA and GaLore. Empirically, MLorc consistently outperforms other memory-efficient training methods, matches or even exceeds the performance of full fine-tuning with a small rank (e.g., $r=4$), and generalizes well across different optimizers -- all while not compromising time or memory efficiency. Furthermore, we provide a theoretical guarantee for its convergence under reasonable assumptions.
- [236] arXiv:2506.01907 [pdf, html, other]
-
Title: SMOTE-DP: Improving Privacy-Utility Tradeoff with Synthetic DataSubjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Machine Learning (stat.ML)
Privacy-preserving data publication, including synthetic data sharing, often experiences trade-offs between privacy and utility. Synthetic data is generally more effective than data anonymization in balancing this trade-off, however, not without its own challenges. Synthetic data produced by generative models trained on source data may inadvertently reveal information about outliers. Techniques specifically designed for preserving privacy, such as introducing noise to satisfy differential privacy, often incur unpredictable and significant losses in utility. In this work we show that, with the right mechanism of synthetic data generation, we can achieve strong privacy protection without significant utility loss. Synthetic data generators producing contracting data patterns, such as Synthetic Minority Over-sampling Technique (SMOTE), can enhance a differentially private data generator, leveraging the strengths of both. We prove in theory and through empirical demonstration that this SMOTE-DP technique can produce synthetic data that not only ensures robust privacy protection but maintains utility in downstream learning tasks.
- [237] arXiv:2506.01913 [pdf, other]
-
Title: Generalized Gradient Norm Clipping & Non-Euclidean $(L_0,L_1)$-SmoothnessSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
This work introduces a hybrid non-Euclidean optimization method which generalizes gradient norm clipping by combining steepest descent and conditional gradient approaches. The method achieves the best of both worlds by establishing a descent property under a generalized notion of ($L_0$,$L_1$)-smoothness. Weight decay is incorporated in a principled manner by identifying a connection to the Frank-Wolfe short step. In the stochastic case, we show an order optimal $O(n^{-1/4})$ convergence rate by leveraging a momentum based gradient estimator. We discuss how to instantiate the algorithms for deep learning and demonstrate their properties on image classification and language modeling.
- [238] arXiv:2506.01919 [pdf, html, other]
-
Title: Transformers as Multi-task Learners: Decoupling Features in Hidden Markov ModelsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Transformer based models have shown remarkable capabilities in sequence learning across a wide range of tasks, often performing well on specific task by leveraging input-output examples. Despite their empirical success, a comprehensive theoretical understanding of this phenomenon remains limited. In this work, we investigate the layerwise behavior of Transformers to uncover the mechanisms underlying their multi-task generalization ability. Taking explorations on a typical sequence model, i.e, Hidden Markov Models, which are fundamental to many language tasks, we observe that: first, lower layers of Transformers focus on extracting feature representations, primarily influenced by neighboring tokens; second, on the upper layers, features become decoupled, exhibiting a high degree of time disentanglement. Building on these empirical insights, we provide theoretical analysis for the expressiveness power of Transformers. Our explicit constructions align closely with empirical observations, providing theoretical support for the Transformer's effectiveness and efficiency on sequence learning across diverse tasks.
New submissions (showing 238 of 238 entries)
- [239] arXiv:2411.19276 (cross-list from quant-ph) [pdf, html, other]
-
Title: Quantum Neural Networks in Practice: A Comparative Study with Classical Models from Standard Data Sets to Industrial ImagesComments: 24 pages, 13 figuresSubjects: Quantum Physics (quant-ph); Machine Learning (cs.LG)
In this study, we compare the performance of randomized classical and quantum neural networks (NNs) as well as classical and quantum-classical hybrid convolutional neural networks (CNNs) for the task of binary image classification. We use two distinct methodologies: using randomized NNs on dimensionality-reduced data, and applying CNNs to full image data. We evaluate these approaches on three data sets of increasing complexity: an artificial hypercube dataset, MNIST handwritten digits and real-world industrial images. We analyze correlations between classification accuracy and quantum model hyperparameters, including the number of trainable parameters, feature encoding methods, circuit layers, entangling gate type and structure, gate entangling power, and measurement operators. For random quantum NNs, we compare their performance against literature models. Classical and quantum/hybrid models achieved statistically equivalent classification accuracies across most datasets, with no approach demonstrating consistent superiority. We observe that quantum models show lower variance with respect to initial training parameters, suggesting better training stability. Among the hyperparameters analyzed, only the number of trainable parameters showed a positive correlation with the model performance. Around 94% of the best-performing quantum NNs had entangling gates, although for hybrid CNNs, models without entanglement performed equally well but took longer to converge. Cross-dataset performance analysis revealed limited transferability of quantum models between different classification tasks. Our study provides an industry perspective on quantum machine learning for practical image classification tasks, highlighting both current limitations and potential avenues for further research in quantum circuit design, entanglement utilization, and model transferability across varied applications.
- [240] arXiv:2506.00002 (cross-list from cs.AR) [pdf, html, other]
-
Title: Advancing AI-assisted Hardware Design with Hierarchical Decentralized Training and Personalized Inference-Time OptimizationSubjects: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
Recent years have witnessed a significant increase in the adoption of AI techniques to enhance electronic design automation. In particular, the emergence of Large Language Models (LLMs) has sparked significant interest in LLM-assisted hardware design generation, spanning applications from classical digital circuits to quantum computing. Despite substantial progress in this direction, the quality of LLM-generated hardware design still cannot meet the requirements for practical deployment. In this work, we identify three critical challenges hindering the development of LLM-assisted hardware design generation: 1) limited data availability, 2) varied data quality, 3) inadequate inference-time efficiency. To address these fundamental challenges, this paper introduces a two-stage framework for AI-assisted hardware design by exploring decentralized training and personalized inference. In the first stage, we propose to harness private domain design sources through a hierarchical decentralized training mechanism that addresses data-sharing constraints. To mitigate the impact of low-quality data, we identify optimization opportunities in hardware generation tasks, using user-defined metrics for model aggregation. The second stage focuses on client personalization to enhance both speed and quality. We introduce a new metric, Trueput, to analyze LLM-assisted hardware generation efficiency. To optimize Trueput, we implement personalized inference-time acceleration and customized sampling strategies. Evaluating both classical and quantum benchmarks, our experimental results demonstrate that the proposed two-stage framework can significantly improve the model capability for hardware design generation. As orthogonal enhancements to existing methods, our framework can achieve $33\% \sim 50\%$ semantic accuracy improvement and $2.3$ times speedup, depending on the difficulty of the generation tasks.
- [241] arXiv:2506.00007 (cross-list from cs.AR) [pdf, other]
-
Title: Emerging ML-AI Techniques for Analog and RF EDAComments: 9 pages, 2 figuresSubjects: Hardware Architecture (cs.AR); Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG)
This survey explores the integration of machine learning (ML) into EDA workflows for analog and RF circuits, addressing challenges unique to analog design, which include complex constraints, nonlinear design spaces, and high computational costs. State-of-the-art learning and optimization techniques are reviewed for circuit tasks such as constraint formulation, topology generation, device modeling, sizing, placement, and routing. The survey highlights the capability of ML to enhance automation, improve design quality, and reduce time-to-market while meeting the target specifications of an analog or RF circuit. Emerging trends and cross-cutting challenges, including robustness to variations and considerations of interconnect parasitics, are also discussed.
- [242] arXiv:2506.00008 (cross-list from cs.AR) [pdf, other]
-
Title: AI Accelerators for Large Language Model In-ference: Architecture Analysis and Scaling StrategiesSubjects: Hardware Architecture (cs.AR); Machine Learning (cs.LG)
The rapid growth of large-language models (LLMs) is driving a new wave of specialized hardware for inference. This paper presents the first workload-centric, cross-architectural performance study of commercial AI accelerators, spanning GPU-based chips, hybrid packages, and wafer-scale engines. We compare memory hierarchies, compute fabrics, and on-chip interconnects, and observe up to 3.7x performance variation across architectures as batch size and sequence length change. Four scaling techniques for trillion-parameter models are examined; expert parallelism offers an 8.4x parameter-to-compute advantage but incurs 2.1x higher latency variance than tensor parallelism. These findings provide quantitative guidance for matching workloads to accelerators and reveal architectural gaps that next-generation designs must address.
- [243] arXiv:2506.00022 (cross-list from cs.CL) [pdf, html, other]
-
Title: Scaling Physical Reasoning with the PHYSICS DatasetShenghe Zheng, Qianjia Cheng, Junchi Yao, Mengsong Wu, haonan he, Ning Ding, Yu Cheng, Shuyue Hu, Lei Bai, Dongzhan Zhou, Ganqu Cui, Peng YeComments: Work on physical datasetsSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG); Physics Education (physics.ed-ph)
Large Language Models (LLMs) have achieved remarkable progress on advanced reasoning tasks such as mathematics and coding competitions. Meanwhile, physics, despite being both reasoning-intensive and essential to real-world understanding, received limited academic and industrial attention. This paper introduces PHYSICS, a dataset containing 16,568 high-quality physics problems spanning subjects and difficulty levels, to facilitate this issue. Specifically, PHYSICS is curated with exercises from over 100 textbooks through a carefully designed pipeline for quality control. It covers five major physics domains: Mechanics, Electromagnetism, Thermodynamics, Optics, and Modern Physics. It also spans a wide range of difficulty levels, from high school to graduate-level physics courses. To utilize the data for improving and evaluating the model's physical reasoning capabilities, we split the dataset into training and test sets, and provide reasoning paths generated by powerful reasoning models for the training data to facilitate model training. In addition, for the evaluation part, we find that existing evaluation frameworks exhibit biases in aspects such as units, simplification, and precision in physics domain. To balance efficiency and accuracy, we introduce a Rule+Model evaluation framework tailored to physics problems. Our evaluations on current state-of-the-art open-source and proprietary models highlight the limitations of current models in handling physics-related tasks. We hope that our dataset and evaluation methodology will jointly advance the development of LLMs in the field of physics.
- [244] arXiv:2506.00033 (cross-list from stat.AP) [pdf, html, other]
-
Title: Probabilistic Spatial Interpolation of Sparse Data using Diffusion ModelsComments: 41 pages, 14 figures, submitted to AMS Artificial Intelligence for the Earth SystemsSubjects: Applications (stat.AP); Machine Learning (cs.LG)
The large underlying assumption of climate models today relies on the basis of a "confident" initial condition, a reasonably plausible snapshot of the Earth for which all future predictions depend on. However, given the inherently chaotic nature of our system, this assumption is complicated by sensitive dependence, where small uncertainties in initial conditions can lead to exponentially diverging outcomes over time. This challenge is particularly salient at global spatial scales and over centennial timescales, where data gaps are not just common but expected. The source of uncertainty is two-fold: (1) sparse, noisy observations from satellites and ground stations, and (2) internal variability stemming from the simplifying approximations within the models themselves.
In practice, data assimilation methods are used to reconcile this missing information by conditioning model states on partial observations. Our work builds on this idea but operates at the extreme end of sparsity. We propose a conditional data imputation framework that reconstructs full temperature fields from as little as 1% observational coverage. The method leverages a diffusion model guided by a prekriged mask, effectively inferring the full-state fields from minimal data points. We validate our framework over the Southern Great Plains, focusing on afternoon (12:00-6:00 PM) temperature fields during the summer months of 2018-2020. Across varying observational densities--from swath data to isolated in-situ sensors--our model achieves strong reconstruction accuracy, highlighting its potential to fill in critical data gaps in both historical reanalysis and real-time forecasting pipelines. - [245] arXiv:2506.00037 (cross-list from cs.IR) [pdf, html, other]
-
Title: Query Drift Compensation: Enabling Compatibility in Continual Learning of Retrieval Embedding ModelsComments: Accepted at CoLLAs 2025Subjects: Information Retrieval (cs.IR); Machine Learning (cs.LG)
Text embedding models enable semantic search, powering several NLP applications like Retrieval Augmented Generation by efficient information retrieval (IR). However, text embedding models are commonly studied in scenarios where the training data is static, thus limiting its applications to dynamic scenarios where new training data emerges over time. IR methods generally encode a huge corpus of documents to low-dimensional embeddings and store them in a database index. During retrieval, a semantic search over the corpus is performed and the document whose embedding is most similar to the query embedding is returned. When updating an embedding model with new training data, using the already indexed corpus is suboptimal due to the non-compatibility issue, since the model which was used to obtain the embeddings of the corpus has changed. While re-indexing of old corpus documents using the updated model enables compatibility, it requires much higher computation and time. Thus, it is critical to study how the already indexed corpus can still be effectively used without the need of re-indexing. In this work, we establish a continual learning benchmark with large-scale datasets and continually train dense retrieval embedding models on query-document pairs from new datasets in each task and observe forgetting on old tasks due to significant drift of embeddings. We employ embedding distillation on both query and document embeddings to maintain stability and propose a novel query drift compensation method during retrieval to project new model query embeddings to the old embedding space. This enables compatibility with previously indexed corpus embeddings extracted using the old model and thus reduces the forgetting. We show that the proposed method significantly improves performance without any re-indexing. Code is available at this https URL.
- [246] arXiv:2506.00041 (cross-list from cs.IR) [pdf, html, other]
-
Title: Decoding Dense Embeddings: Sparse Autoencoders for Interpreting and Discretizing Dense RetrievalSubjects: Information Retrieval (cs.IR); Machine Learning (cs.LG)
Despite their strong performance, Dense Passage Retrieval (DPR) models suffer from a lack of interpretability. In this work, we propose a novel interpretability framework that leverages Sparse Autoencoders (SAEs) to decompose previously uninterpretable dense embeddings from DPR models into distinct, interpretable latent concepts. We generate natural language descriptions for each latent concept, enabling human interpretations of both the dense embeddings and the query-document similarity scores of DPR models. We further introduce Concept-Level Sparse Retrieval (CL-SR), a retrieval framework that directly utilizes the extracted latent concepts as indexing units. CL-SR effectively combines the semantic expressiveness of dense embeddings with the transparency and efficiency of sparse representations. We show that CL-SR achieves high index-space and computational efficiency while maintaining robust performance across vocabulary and semantic mismatches.
- [247] arXiv:2506.00044 (cross-list from stat.AP) [pdf, html, other]
-
Title: Probabilistic intraday electricity price forecasting using generative machine learningSubjects: Applications (stat.AP); Machine Learning (cs.LG); Machine Learning (stat.ML)
The growing importance of intraday electricity trading in Europe calls for improved price forecasting and tailored decision-support tools. In this paper, we propose a novel generative neural network model to generate probabilistic path forecasts for intraday electricity prices and use them to construct effective trading strategies for Germany's continuous-time intraday market. Our method demonstrates competitive performance in terms of statistical evaluation metrics compared to two state-of-the-art statistical benchmark approaches. To further assess its economic value, we consider a realistic fixed-volume trading scenario and propose various strategies for placing market sell orders based on the path forecasts. Among the different trading strategies, the price paths generated by our generative model lead to higher profit gains than the benchmark methods. Our findings highlight the potential of generative machine learning tools in electricity price forecasting and underscore the importance of economic evaluation.
- [248] arXiv:2506.00048 (cross-list from cs.IR) [pdf, html, other]
-
Title: Graph Contrastive Learning for Optimizing Sparse Data in Recommender Systems with LightGCLComments: Term Paper, Machine Learning with Graphs, North Carolina State UniversitySubjects: Information Retrieval (cs.IR); Machine Learning (cs.LG)
Graph Neural Networks (GNNs) are powerful tools for recommendation systems, but they often struggle under data sparsity and noise. To address these issues, we implemented LightGCL, a graph contrastive learning model that uses Singular Value Decomposition (SVD) for robust graph augmentation, preserving semantic integrity without relying on stochastic or heuristic perturbations. LightGCL enables structural refinement and captures global collaborative signals, achieving significant gains over state-of-the-art models across benchmark datasets. Our experiments also demonstrate improved fairness and resilience to popularity bias, making it well-suited for real-world recommender systems.
- [249] arXiv:2506.00053 (cross-list from q-bio.QM) [pdf, other]
-
Title: Improving statistical learning methods via features selection without replacement sampling and random projectionSubjects: Quantitative Methods (q-bio.QM); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Applications (stat.AP); Machine Learning (stat.ML)
Cancer is fundamentally a genetic disease characterized by genetic and epigenetic alterations that disrupt normal gene expression, leading to uncontrolled cell growth and metastasis. High-dimensional microarray datasets pose challenges for classification models due to the "small n, large p" problem, resulting in overfitting. This study makes three different key contributions: 1) we propose a machine learning-based approach integrating the Feature Selection Without Re-placement (FSWOR) technique and a projection method to improve classification accuracy. 2) We apply the Kendall statistical test to identify the most significant genes from the brain cancer mi-croarray dataset (GSE50161), reducing the feature space from 54,675 to 20,890 genes.3) we apply machine learning models using k-fold cross validation techniques in which our model incorpo-rates ensemble classifiers with LDA projection and Naïve Bayes, achieving a test score of 96%, outperforming existing methods by 9.09%. The results demonstrate the effectiveness of our ap-proach in high-dimensional gene expression analysis, improving classification accuracy while mitigating overfitting. This study contributes to cancer biomarker discovery, offering a robust computational method for analyzing microarray data.
- [250] arXiv:2506.00057 (cross-list from cs.CY) [pdf, html, other]
-
Title: Hierarchical Bayesian Knowledge Tracing in Undergraduate Engineering EducationComments: 6 pages, 6 figures, 3 tablesSubjects: Computers and Society (cs.CY); Machine Learning (cs.LG); Applications (stat.AP); Machine Learning (stat.ML)
Educators teaching entry-level university engineering modules face the challenge of identifying which topics students find most difficult and how to support diverse student needs effectively. This study demonstrates a rigorous yet interpretable statistical approach -- hierarchical Bayesian modeling -- that leverages detailed student response data to quantify both skill difficulty and individual student abilities. Using a large-scale dataset from an undergraduate Statics course, we identified clear patterns of skill mastery and uncovered distinct student subgroups based on their learning trajectories. Our analysis reveals that certain concepts consistently present challenges, requiring targeted instructional support, while others are readily mastered and may benefit from enrichment activities. Importantly, the hierarchical Bayesian method provides educators with intuitive, reliable metrics without sacrificing predictive accuracy. This approach allows for data-informed decisions, enabling personalized teaching strategies to improve student engagement and success. By combining robust statistical methods with clear interpretability, this study equips educators with actionable insights to better support diverse learner populations.
- [251] arXiv:2506.00062 (cross-list from cs.CY) [pdf, html, other]
-
Title: SafeCOMM: What about Safety Alignment in Fine-Tuned Telecom Large Language Models?Subjects: Computers and Society (cs.CY); Computation and Language (cs.CL); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
Fine-tuning large language models (LLMs) for telecom tasks and datasets is a common practice to adapt general-purpose models to the telecom domain. However, little attention has been paid to how this process may compromise model safety. Recent research has shown that even benign fine-tuning can degrade the safety alignment of LLMs, causing them to respond to harmful or unethical user queries. In this paper, we investigate this issue for telecom-tuned LLMs using three representative datasets featured by the GenAINet initiative. We show that safety degradation persists even for structured and seemingly harmless datasets such as 3GPP standards and tabular records, indicating that telecom-specific data is not immune to safety erosion during fine-tuning. We further extend our analysis to publicly available Telecom LLMs trained via continual pre-training, revealing that safety alignment is often severely lacking, primarily due to the omission of safety-focused instruction tuning. To address these issues in both fine-tuned and pre-trained models, we conduct extensive experiments and evaluate three safety realignment defenses (SafeInstruct, SafeLoRA, and SafeMERGE) using established red-teaming benchmarks. The results show that, across all settings, the proposed defenses can effectively restore safety after harmful degradation without compromising downstream task performance, leading to Safe teleCOMMunication (SafeCOMM) models. In a nutshell, our work serves as a diagnostic study and practical guide for safety realignment in telecom-tuned LLMs, and emphasizes the importance of safety-aware instruction and fine-tuning for real-world deployments of Telecom LLMs.
- [252] arXiv:2506.00072 (cross-list from cs.CY) [pdf, other]
-
Title: Evaluating Prompt Engineering Techniques for Accuracy and Confidence Elicitation in Medical LLMsNariman Naderi, Zahra Atf, Peter R Lewis, Aref Mahjoub far, Seyed Amir Ahmad Safavi-Naini, Ali SoroushComments: This paper was accepted for presentation at the 7th International Workshop on EXplainable, Trustworthy, and Responsible AI and Multi-Agent Systems (EXTRAAMAS 2025). Workshop website: this https URLSubjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
This paper investigates how prompt engineering techniques impact both accuracy and confidence elicitation in Large Language Models (LLMs) applied to medical contexts. Using a stratified dataset of Persian board exam questions across multiple specialties, we evaluated five LLMs - GPT-4o, o3-mini, Llama-3.3-70b, Llama-3.1-8b, and DeepSeek-v3 - across 156 configurations. These configurations varied in temperature settings (0.3, 0.7, 1.0), prompt styles (Chain-of-Thought, Few-Shot, Emotional, Expert Mimicry), and confidence scales (1-10, 1-100). We used AUC-ROC, Brier Score, and Expected Calibration Error (ECE) to evaluate alignment between confidence and actual performance. Chain-of-Thought prompts improved accuracy but also led to overconfidence, highlighting the need for calibration. Emotional prompting further inflated confidence, risking poor decisions. Smaller models like Llama-3.1-8b underperformed across all metrics, while proprietary models showed higher accuracy but still lacked calibrated confidence. These results suggest prompt engineering must address both accuracy and uncertainty to be effective in high-stakes medical tasks.
- [253] arXiv:2506.00076 (cross-list from cs.CY) [pdf, other]
-
Title: Optimizing Storytelling, Improving Audience Retention, and Reducing Waste in the Entertainment IndustrySubjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Television networks face high financial risk when making programming decisions, often relying on limited historical data to forecast episodic viewership. This study introduces a machine learning framework that integrates natural language processing (NLP) features from over 25000 television episodes with traditional viewership data to enhance predictive accuracy. By extracting emotional tone, cognitive complexity, and narrative structure from episode dialogue, we evaluate forecasting performance using SARIMAX, rolling XGBoost, and feature selection models. While prior viewership remains a strong baseline predictor, NLP features contribute meaningful improvements for some series. We also introduce a similarity scoring method based on Euclidean distance between aggregate dialogue vectors to compare shows by content. Tested across diverse genres, including Better Call Saul and Abbott Elementary, our framework reveals genre-specific performance and offers interpretable metrics for writers, executives, and marketers seeking data-driven insight into audience behavior.
- [254] arXiv:2506.00077 (cross-list from cs.CL) [pdf, html, other]
-
Title: Gaussian mixture models as a proxy for interacting language modelsSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG); Machine Learning (stat.ML)
Large language models (LLMs) are a powerful tool with the ability to match human capabilities and behavior in many settings. Retrieval-augmented generation (RAG) further allows LLMs to generate diverse output depending on the contents of their RAG database. This motivates their use in the social sciences to study human behavior between individuals when large-scale experiments are infeasible. However, LLMs depend on complex, computationally expensive algorithms. In this paper, we introduce interacting Gaussian mixture models (GMMs) as an alternative to similar frameworks using LLMs. We compare a simplified model of GMMs to select experimental simulations of LLMs whose updating and response depend on feedback from other LLMs. We find that interacting GMMs capture important features of the dynamics in interacting LLMs, and we investigate key similarities and differences between interacting LLMs and GMMs. We conclude by discussing the benefits of Gaussian mixture models, potential modifications, and future research directions.
- [255] arXiv:2506.00079 (cross-list from cs.CY) [pdf, html, other]
-
Title: Who Gets the Kidney? Human-AI Alignment, Indecision, and Moral ValuesSubjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
The rapid integration of Large Language Models (LLMs) in high-stakes decision-making -- such as allocating scarce resources like donor organs -- raises critical questions about their alignment with human moral values. We systematically evaluate the behavior of several prominent LLMs against human preferences in kidney allocation scenarios and show that LLMs: i) exhibit stark deviations from human values in prioritizing various attributes, and ii) in contrast to humans, LLMs rarely express indecision, opting for deterministic decisions even when alternative indecision mechanisms (e.g., coin flipping) are provided. Nonetheless, we show that low-rank supervised fine-tuning with few samples is often effective in improving both decision consistency and calibrating indecision modeling. These findings illustrate the necessity of explicit alignment strategies for LLMs in moral/ethical domains.
- [256] arXiv:2506.00088 (cross-list from cs.CL) [pdf, html, other]
-
Title: HD-NDEs: Neural Differential Equations for Hallucination Detection in LLMsSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
In recent years, large language models (LLMs) have made remarkable advancements, yet hallucination, where models produce inaccurate or non-factual statements, remains a significant challenge for real-world deployment. Although current classification-based methods, such as SAPLMA, are highly efficient in mitigating hallucinations, they struggle when non-factual information arises in the early or mid-sequence of outputs, reducing their reliability. To address these issues, we propose Hallucination Detection-Neural Differential Equations (HD-NDEs), a novel method that systematically assesses the truthfulness of statements by capturing the full dynamics of LLMs within their latent space. Our approaches apply neural differential equations (Neural DEs) to model the dynamic system in the latent space of LLMs. Then, the sequence in the latent space is mapped to the classification space for truth assessment. The extensive experiments across five datasets and six widely used LLMs demonstrate the effectiveness of HD-NDEs, especially, achieving over 14% improvement in AUC-ROC on the True-False dataset compared to state-of-the-art techniques.
- [257] arXiv:2506.00098 (cross-list from cs.RO) [pdf, other]
-
Title: Interactive Imitation Learning for Dexterous Robotic Manipulation: Challenges and Perspectives -- A SurveyComments: 21 pages, 3 figuresSubjects: Robotics (cs.RO); Machine Learning (cs.LG)
Dexterous manipulation is a crucial yet highly complex challenge in humanoid robotics, demanding precise, adaptable, and sample-efficient learning methods. As humanoid robots are usually designed to operate in human-centric environments and interact with everyday objects, mastering dexterous manipulation is critical for real-world deployment. Traditional approaches, such as reinforcement learning and imitation learning, have made significant strides, but they often struggle due to the unique challenges of real-world dexterous manipulation, including high-dimensional control, limited training data, and covariate shift. This survey provides a comprehensive overview of these challenges and reviews existing learning-based methods for dexterous manipulation, spanning imitation learning, reinforcement learning, and hybrid approaches. A promising yet underexplored direction is interactive imitation learning, where human feedback actively refines a robot's behavior during training. While interactive imitation learning has shown success in various robotic tasks, its application to dexterous manipulation remains limited. To address this gap, we examine current interactive imitation learning techniques applied to other robotic tasks and discuss how these methods can be adapted to enhance dexterous manipulation. By synthesizing state-of-the-art research, this paper highlights key challenges, identifies gaps in current methodologies, and outlines potential directions for leveraging interactive imitation learning to improve dexterous robotic skills.
- [258] arXiv:2506.00102 (cross-list from hep-ph) [pdf, html, other]
-
Title: Tensor Network for Anomaly Detection in the Latent Space of Proton Collision Events at the LHCSubjects: High Energy Physics - Phenomenology (hep-ph); Statistical Mechanics (cond-mat.stat-mech); Machine Learning (cs.LG); High Energy Physics - Experiment (hep-ex); Quantum Physics (quant-ph); Machine Learning (stat.ML)
The pursuit of discovering new phenomena at the Large Hadron Collider (LHC) demands constant innovation in algorithms and technologies. Tensor networks are mathematical models on the intersection of classical and quantum machine learning, which present a promising and efficient alternative for tackling these challenges. In this work, we propose a tensor network-based strategy for anomaly detection at the LHC and demonstrate its superior performance in identifying new phenomena compared to established quantum methods. Our model is a parametrized Matrix Product State with an isometric feature map, processing a latent representation of simulated LHC data generated by an autoencoder. Our results highlight the potential of tensor networks to enhance new-physics discovery.
- [259] arXiv:2506.00119 (cross-list from hep-ph) [pdf, html, other]
-
Title: Generator Based Inference (GBI)Chi Lung Cheng, Ranit Das, Runze Li, Radha Mastandrea, Vinicius Mikuni, Benjamin Nachman, David Shih, Gup SinghComments: 9 pages, 9 figuresSubjects: High Energy Physics - Phenomenology (hep-ph); Machine Learning (cs.LG); High Energy Physics - Experiment (hep-ex)
Statistical inference in physics is often based on samples from a generator (sometimes referred to as a ``forward model") that emulate experimental data and depend on parameters of the underlying theory. Modern machine learning has supercharged this workflow to enable high-dimensional and unbinned analyses to utilize much more information than ever before. We propose a general framework for describing the integration of machine learning with generators called Generator Based Inference (GBI). A well-studied special case of this setup is Simulation Based Inference (SBI) where the generator is a physics-based simulator. In this work, we examine other methods within the GBI toolkit that use data-driven methods to build the generator. In particular, we focus on resonant anomaly detection, where the generator describing the background is learned from sidebands. We show how to perform machine learning-based parameter estimation in this context with data-derived generators. This transforms the statistical outputs of anomaly detection to be directly interpretable and the performance on the LHCO community benchmark dataset establishes a new state-of-the-art for anomaly detection sensitivity.
- [260] arXiv:2506.00128 (cross-list from cs.SE) [pdf, html, other]
-
Title: Applying Large Language Models to Issue Classification: Revisiting with Extended Data and New ModelsComments: 35 pages, 2 figures, 9 tables, Pre-print for Science of Computer ProgrammingSubjects: Software Engineering (cs.SE); Machine Learning (cs.LG)
Effective prioritization of issue reports in software engineering helps to optimize resource allocation and information recovery. However, manual issue classification is laborious and lacks scalability. As an alternative, many open source software (OSS) projects employ automated processes for this task, yet this method often relies on large datasets for adequate training. Traditionally, machine learning techniques have been used for issue classification. More recently, large language models (LLMs) have emerged as powerful tools for addressing a range of software engineering challenges, including code and test generation, mapping new requirements to legacy software endpoints, and conducting code reviews. The following research investigates an automated approach to issue classification based on LLMs. By leveraging the capabilities of such models, we aim to develop a robust system for prioritizing issue reports, mitigating the necessity for extensive training data while also maintaining reliability in classification. In our research, we developed an LLM-based approach for accurately labeling issues by selecting two of the most prominent large language models. We then compared their performance across multiple datasets. Our findings show that GPT-4o achieved the best results in classifying issues from the NLBSE 2024 competition. Moreover, GPT-4o outperformed DeepSeek R1, achieving an F1 score 20% higher when both models were trained on the same dataset from the NLBSE 2023 competition, which was ten times larger than the NLBSE 2024 dataset. The fine-tuned GPT-4o model attained an average F1 score of 80.7%, while the fine-tuned DeepSeek R1 model achieved 59.33%. Increasing the dataset size did not improve the F1 score, reducing the dependence on massive datasets for building an efficient solution to issue classification.
- [261] arXiv:2506.00129 (cross-list from cs.CV) [pdf, html, other]
-
Title: Geo-Sign: Hyperbolic Contrastive Regularisation for Geometrically Aware Sign Language TranslationComments: Under ReviewSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Recent progress in Sign Language Translation (SLT) has focussed primarily on improving the representational capacity of large language models to incorporate Sign Language features. This work explores an alternative direction: enhancing the geometric properties of skeletal representations themselves. We propose Geo-Sign, a method that leverages the properties of hyperbolic geometry to model the hierarchical structure inherent in sign language kinematics. By projecting skeletal features derived from Spatio-Temporal Graph Convolutional Networks (ST-GCNs) into the Poincaré ball model, we aim to create more discriminative embeddings, particularly for fine-grained motions like finger articulations. We introduce a hyperbolic projection layer, a weighted Fréchet mean aggregation scheme, and a geometric contrastive loss operating directly in hyperbolic space. These components are integrated into an end-to-end translation framework as a regularisation function, to enhance the representations within the language model. This work demonstrates the potential of hyperbolic geometry to improve skeletal representations for Sign Language Translation, improving on SOTA RGB methods while preserving privacy and improving computational efficiency. Code available here: this https URL.
- [262] arXiv:2506.00133 (cross-list from cs.NI) [pdf, html, other]
-
Title: A Reinforcement Learning-Based Telematic Routing Protocol for the Internet of Underwater ThingsComments: 8 Pages, 10 Figures, 2 TablesSubjects: Networking and Internet Architecture (cs.NI); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
The Internet of Underwater Things (IoUT) faces major challenges such as low bandwidth, high latency, mobility, and limited energy resources. Traditional routing protocols like RPL, which were designed for land-based networks, do not perform well in these underwater conditions. This paper introduces RL-RPL-UA, a new routing protocol that uses reinforcement learning to improve performance in underwater environments. Each node includes a lightweight RL agent that selects the best parent node based on local information such as packet delivery ratio, buffer level, link quality, and remaining energy. RL-RPL-UA keeps full compatibility with standard RPL messages and adds a dynamic objective function to support real-time decision-making. Simulations using Aqua-Sim show that RL-RPL-UA increases packet delivery by up to 9.2%, reduces energy use per packet by 14.8%, and extends network lifetime by 80 seconds compared to traditional methods. These results suggest that RL-RPL-UA is a promising and energy-efficient routing solution for underwater networks.
- [263] arXiv:2506.00137 (cross-list from cs.CL) [pdf, html, other]
-
Title: LaMP-QA: A Benchmark for Personalized Long-form Question AnsweringSubjects: Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
Personalization is essential for question answering systems that are user-centric. Despite its importance, personalization in answer generation has been relatively underexplored. This is mainly due to lack of resources for training and evaluating personalized question answering systems. We address this gap by introducing LaMP-QA -- a benchmark designed for evaluating personalized long-form answer generation. The benchmark covers questions from three major categories: (1) Arts & Entertainment, (2) Lifestyle & Personal Development, and (3) Society & Culture, encompassing over 45 subcategories in total. To assess the quality and potential impact of the LaMP-QA benchmark for personalized question answering, we conduct comprehensive human and automatic evaluations, to compare multiple evaluation strategies for evaluating generated personalized responses and measure their alignment with human preferences. Furthermore, we benchmark a number of non-personalized and personalized approaches based on open-source and proprietary large language models (LLMs). Our results show that incorporating the personalized context provided leads to performance improvements of up to 39%. The benchmark is publicly released to support future research in this area.
- [264] arXiv:2506.00138 (cross-list from q-bio.NC) [pdf, html, other]
-
Title: Autonomous Behavior and Whole-Brain Dynamics Emerge in Embodied Zebrafish Agents with Model-based Intrinsic MotivationComments: 17 pages, 7 figuresSubjects: Neurons and Cognition (q-bio.NC); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Robotics (cs.RO)
Autonomy is a hallmark of animal intelligence, enabling adaptive and intelligent behavior in complex environments without relying on external reward or task structure. Existing reinforcement learning approaches to exploration in sparse reward and reward-free environments, including class of methods known as intrinsic motivation, exhibit inconsistent exploration patterns and thus fail to produce robust autonomous behaviors observed in animals. Moreover, systems neuroscience has largely overlooked the neural basis of autonomy, focusing instead on experimental paradigms where animals are motivated by external reward rather than engaging in unconstrained, naturalistic and task-independent behavior. To bridge these gaps, we introduce a novel model-based intrinsic drive explicitly designed to capture robust autonomous exploration observed in animals. Our method (3M-Progress) motivates naturalistic behavior by tracking divergence between the agent's current world model and an ethological prior. We demonstrate that artificial embodied agents trained with 3M-Progress capture the explainable variance in behavioral patterns and whole-brain neural-glial dynamics recorded from autonomously-behaving larval zebrafish, introducing the first goal-driven, population-level model of neural-glial computation. Our findings establish a computational framework connecting model-based intrinsic motivation to naturalistic behavior, providing a foundation for building artificial agents with animal-like autonomy.
- [265] arXiv:2506.00140 (cross-list from cs.AI) [pdf, html, other]
-
Title: Balancing Profit and Fairness in Risk-Based Pricing MarketsSubjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); General Economics (econ.GN)
Dynamic, risk-based pricing can systematically exclude vulnerable consumer groups from essential resources such as health insurance and consumer credit. We show that a regulator can realign private incentives with social objectives through a learned, interpretable tax schedule. First, we provide a formal proposition that bounding each firm's \emph{local} demographic gap implicitly bounds the \emph{global} opt-out disparity, motivating firm-level penalties. Building on this insight we introduce \texttt{MarketSim} -- an open-source, scalable simulator of heterogeneous consumers and profit-maximizing firms -- and train a reinforcement learning (RL) social planner (SP) that selects a bracketed fairness-tax while remaining close to a simple linear prior via an $\mathcal{L}_1$ regularizer. The learned policy is thus both transparent and easily interpretable. In two empirically calibrated markets, i.e., U.S. health-insurance and consumer-credit, our planner simultaneously raises demand-fairness by up to $16\%$ relative to unregulated Free Market while outperforming a fixed linear schedule in terms of social welfare without explicit coordination. These results illustrate how AI-assisted regulation can convert a competitive social dilemma into a win-win equilibrium, providing a principled and practical framework for fairness-aware market oversight.
- [266] arXiv:2506.00165 (cross-list from cs.DS) [pdf, html, other]
-
Title: Randomized Dimensionality Reduction for Euclidean Maximization and Diversity MeasuresJie Gao, Rajesh Jayaram, Benedikt Kolbe, Shay Sapir, Chris Schwiegelshohn, Sandeep Silwal, Erik WaingartenComments: ICML 2025Subjects: Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG)
Randomized dimensionality reduction is a widely-used algorithmic technique for speeding up large-scale Euclidean optimization problems. In this paper, we study dimension reduction for a variety of maximization problems, including max-matching, max-spanning tree, max TSP, as well as various measures for dataset diversity. For these problems, we show that the effect of dimension reduction is intimately tied to the \emph{doubling dimension} $\lambda_X$ of the underlying dataset $X$ -- a quantity measuring intrinsic dimensionality of point sets. Specifically, we prove that a target dimension of $O(\lambda_X)$ suffices to approximately preserve the value of any near-optimal solution,which we also show is necessary for some of these problems. This is in contrast to classical dimension reduction results, whose dependence increases with the dataset size $|X|$. We also provide empirical results validating the quality of solutions found in the projected space, as well as speedups due to dimensionality reduction.
- [267] arXiv:2506.00171 (cross-list from stat.ML) [pdf, other]
-
Title: Minimax Rates for the Estimation of Eigenpairs of Weighted Laplace-Beltrami Operators on ManifoldsSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Analysis of PDEs (math.AP)
We study the problem of estimating eigenpairs of elliptic differential operators from samples of a distribution $\rho$ supported on a manifold $M$. The operators discussed in the paper are relevant in unsupervised learning and in particular are obtained by taking suitable scaling limits of widely used graph Laplacians over data clouds. We study the minimax risk for this eigenpair estimation problem and explore the rates of approximation that can be achieved by commonly used graph Laplacians built from random data. More concretely, assuming that $\rho$ belongs to a certain family of distributions with controlled second derivatives, and assuming that the $d$-dimensional manifold $M$ where $\rho$ is supported has bounded geometry, we prove that the statistical minimax rate for approximating eigenvalues and eigenvectors in the $H^1(M)$-sense is $n^{-2/(d+4)}$, a rate that matches the minimax rate for a closely related density estimation problem. We then revisit the literature studying Laplacians over proximity graphs in the large data limit and prove that, under slightly stronger regularity assumptions on the data generating model, eigenpairs of graph Laplacians induce manifold agnostic estimators with an error of approximation that, up to logarithmic corrections, matches our lower bounds. Our analysis allows us to expand the existing literature on graph-based learning in at least two significant ways: 1) we consider stronger norms to measure the error of approximation than the ones that had been analyzed in the past; 2) our rates of convergence are uniform over a family of smooth distributions and do not just apply to densities with special symmetries, and, as a consequence of our lower bounds, are essentially sharp when the connectivity of the graph is sufficiently high.
- [268] arXiv:2506.00180 (cross-list from cs.GT) [pdf, html, other]
-
Title: Empirical Validation of the Independent Chip ModelSubjects: Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG)
The independent chip model (ICM) forms a cornerstone of all modern poker tournament strategy. However, despite its prominence, the ICM's performance in the real world has not been sufficiently scrutinized, especially at a large scale. In this paper, we introduce our new dataset of poker tournaments, consisting of results of over ten thousand events. Then, using this dataset, we perform two experiments as part of a large-scale empirical validation of the ICM. First, we verify that the ICM performs more accurately than a baseline we propose. Second, we obtain empirical evidence of the ICM underestimating the performances of players with larger stacks while overestimating those who are short-stacked. Our contributions may be useful to future researchers developing new algorithms for estimating a player's value in poker tournaments.
- [269] arXiv:2506.00182 (cross-list from stat.ML) [pdf, other]
-
Title: Overfitting has a limitation: a model-independent generalization error bound based on Rényi entropySubjects: Machine Learning (stat.ML); Information Theory (cs.IT); Machine Learning (cs.LG); Statistics Theory (math.ST)
Will further scaling up of machine learning models continue to bring success? A significant challenge in answering this question lies in understanding generalization error, which is the impact of overfitting. Understanding generalization error behavior of increasingly large-scale machine learning models remains a significant area of investigation, as conventional analyses often link error bounds to model complexity, failing to fully explain the success of extremely large architectures. This research introduces a novel perspective by establishing a model-independent upper bound for generalization error applicable to algorithms whose outputs are determined solely by the data's histogram, such as empirical risk minimization or gradient-based methods. Crucially, this bound is shown to depend only on the Rényi entropy of the data-generating distribution, suggesting that a small generalization error can be maintained even with arbitrarily large models, provided the data quantity is sufficient relative to this entropy. This framework offers a direct explanation for the phenomenon where generalization performance degrades significantly upon injecting random noise into data, where the performance degrade is attributed to the consequent increase in the data distribution's Rényi entropy. Furthermore, we adapt the no-free-lunch theorem to be data-distribution-dependent, demonstrating that an amount of data corresponding to the Rényi entropy is indeed essential for successful learning, thereby highlighting the tightness of our proposed generalization bound.
- [270] arXiv:2506.00185 (cross-list from eess.AS) [pdf, html, other]
-
Title: Pushing the Limits of Beam Search Decoding for Transducer-based ASR modelsComments: Accepted to Interspeech 2025Subjects: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Sound (cs.SD)
Transducer models have emerged as a promising choice for end-to-end ASR systems, offering a balanced trade-off between recognition accuracy, streaming capabilities, and inference speed in greedy decoding. However, beam search significantly slows down Transducers due to repeated evaluations of key network components, limiting practical applications. This paper introduces a universal method to accelerate beam search for Transducers, enabling the implementation of two optimized algorithms: ALSD++ and AES++. The proposed method utilizes batch operations, a tree-based hypothesis structure, novel blank scoring for enhanced shallow fusion, and CUDA graph execution for efficient GPU inference. This narrows the speed gap between beam and greedy modes to only 10-20% for the whole system, achieves 14-30% relative improvement in WER compared to greedy decoding, and improves shallow fusion for low-resource up to 11% compared to existing implementations. All the algorithms are open sourced.
- [271] arXiv:2506.00191 (cross-list from cs.CR) [pdf, html, other]
-
Title: Heterogeneous Graph Backdoor AttackSubjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Heterogeneous Graph Neural Networks (HGNNs) excel in modeling complex, multi-typed relationships across diverse domains, yet their vulnerability to backdoor attacks remains unexplored. To address this gap, we conduct the first investigation into the susceptibility of HGNNs to existing graph backdoor attacks, revealing three critical issues: (1) high attack budget required for effective backdoor injection, (2) inefficient and unreliable backdoor activation, and (3) inaccurate attack effectiveness evaluation. To tackle these issues, we propose the Heterogeneous Graph Backdoor Attack (HGBA), the first backdoor attack specifically designed for HGNNs, introducing a novel relation-based trigger mechanism that establishes specific connections between a strategically selected trigger node and poisoned nodes via the backdoor metapath. HGBA achieves efficient and stealthy backdoor injection with minimal structural modifications and supports easy backdoor activation through two flexible strategies: Self-Node Attack and Indiscriminate Attack. Additionally, we improve the ASR measurement protocol, enabling a more accurate assessment of attack effectiveness. Extensive experiments demonstrate that HGBA far surpasses multiple state-of-the-art graph backdoor attacks in black-box settings, efficiently attacking HGNNs with low attack budgets. Ablation studies show that the strength of HBGA benefits from our trigger node selection method and backdoor metapath selection strategy. In addition, HGBA shows superior robustness against node feature perturbations and multiple types of existing graph backdoor defense mechanisms. Finally, extension experiments demonstrate that the relation-based trigger mechanism can effectively extend to tasks in homogeneous graph scenarios, thereby posing severe threats to broader security-critical domains.
- [272] arXiv:2506.00197 (cross-list from cs.CR) [pdf, html, other]
-
Title: When GPT Spills the Tea: Comprehensive Assessment of Knowledge File Leakage in GPTsSubjects: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
Knowledge files have been widely used in large language model (LLM) agents, such as GPTs, to improve response quality. However, concerns about the potential leakage of knowledge files have grown significantly. Existing studies demonstrate that adversarial prompts can induce GPTs to leak knowledge file content. Yet, it remains uncertain whether additional leakage vectors exist, particularly given the complex data flows across clients, servers, and databases in GPTs. In this paper, we present a comprehensive risk assessment of knowledge file leakage, leveraging a novel workflow inspired by Data Security Posture Management (DSPM). Through the analysis of 651,022 GPT metadata, 11,820 flows, and 1,466 responses, we identify five leakage vectors: metadata, GPT initialization, retrieval, sandboxed execution environments, and prompts. These vectors enable adversaries to extract sensitive knowledge file data such as titles, content, types, and sizes. Notably, the activation of the built-in tool Code Interpreter leads to a privilege escalation vulnerability, enabling adversaries to directly download original knowledge files with a 95.95% success rate. Further analysis reveals that 28.80% of leaked files are copyrighted, including digital copies from major publishers and internal materials from a listed company. In the end, we provide actionable solutions for GPT builders and platform providers to secure the GPT data supply chain.
- [273] arXiv:2506.00200 (cross-list from cs.CL) [pdf, html, other]
-
Title: Structuring Radiology Reports: Challenging LLMs with Lightweight ModelsJohannes Moll, Louisa Fay, Asfandyar Azhar, Sophie Ostmeier, Tim Lueth, Sergios Gatidis, Curtis Langlotz, Jean-Benoit DelbrouckSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Radiology reports are critical for clinical decision-making but often lack a standardized format, limiting both human interpretability and machine learning (ML) applications. While large language models (LLMs) have shown strong capabilities in reformatting clinical text, their high computational requirements, lack of transparency, and data privacy concerns hinder practical deployment. To address these challenges, we explore lightweight encoder-decoder models (<300M parameters)-specifically T5 and BERT2BERT-for structuring radiology reports from the MIMIC-CXR and CheXpert Plus datasets. We benchmark these models against eight open-source LLMs (1B-70B), adapted using prefix prompting, in-context learning (ICL), and low-rank adaptation (LoRA) finetuning. Our best-performing lightweight model outperforms all LLMs adapted using prompt-based techniques on a human-annotated test set. While some LoRA-finetuned LLMs achieve modest gains over the lightweight model on the Findings section (BLEU 6.4%, ROUGE-L 4.8%, BERTScore 3.6%, F1-RadGraph 1.1%, GREEN 3.6%, and F1-SRR-BERT 4.3%), these improvements come at the cost of substantially greater computational resources. For example, LLaMA-3-70B incurred more than 400 times the inference time, cost, and carbon emissions compared to the lightweight model. These results underscore the potential of lightweight, task-specific models as sustainable and privacy-preserving solutions for structuring clinical text in resource-constrained healthcare settings.
- [274] arXiv:2506.00223 (cross-list from q-bio.QM) [pdf, html, other]
-
Title: Enhancing Drug Discovery: Autoencoder-Based Latent Space Augmentation for Improved Molecular Solubility Prediction using LatMixSolSubjects: Quantitative Methods (q-bio.QM); Machine Learning (cs.LG)
Accurate prediction of molecular solubility is a cornerstone of early-stage drug discovery, yet conventional machine learning models face significant challenges due to limited labeled data and the high-dimensional nature of molecular descriptors. To address these issues, we propose LatMixSol, a novel latent space augmentation framework that combines autoencoder-based feature compression with guided interpolation to enrich training data. Our approach first encodes molecular descriptors into a low-dimensional latent space using a two-layer autoencoder. Spectral clustering is then applied to group chemically similar molecules, enabling targeted MixUp-style interpolation within clusters. Synthetic samples are generated by blending latent vectors of cluster members and decoding them back to the original feature space. Evaluated on the Huuskonen solubility benchmark, LatMixSol demonstrates consistent improvements across three of four gradient-boosted regressors (CatBoost, LightGBM, HistGradientBoosting), achieving RMSE reductions of 3.2-7.6% and R-squared increases of 0.5-1.5%. Notably, HistGradientBoosting shows the most significant enhancement with a 7.6% RMSE improvement. Our analysis confirms that cluster-guided latent space augmentation preserves chemical validity while expanding dataset diversity, offering a computationally efficient strategy to enhance predictive models in resource-constrained drug discovery pipelines.
- [275] arXiv:2506.00226 (cross-list from stat.ML) [pdf, html, other]
-
Title: Riemannian Principal Component AnalysisSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST); Computation (stat.CO)
This paper proposes an innovative extension of Principal Component Analysis (PCA) that transcends the traditional assumption of data lying in Euclidean space, enabling its application to data on Riemannian manifolds. The primary challenge addressed is the lack of vector space operations on such manifolds. Fletcher et al., in their work {\em Principal Geodesic Analysis for the Study of Nonlinear Statistics of Shape}, proposed Principal Geodesic Analysis (PGA) as a geometric approach to analyze data on Riemannian manifolds, particularly effective for structured datasets like medical images, where the manifold's intrinsic structure is apparent. However, PGA's applicability is limited when dealing with general datasets that lack an implicit local distance notion. In this work, we introduce a generalized framework, termed {\em Riemannian Principal Component Analysis (R-PCA)}, to extend PGA for any data endowed with a local distance structure. Specifically, we adapt the PCA methodology to Riemannian manifolds by equipping data tables with local metrics, enabling the incorporation of manifold geometry. This framework provides a unified approach for dimensionality reduction and statistical analysis directly on manifolds, opening new possibilities for datasets with region-specific or part-specific distance notions, ensuring respect for their intrinsic geometric properties.
- [276] arXiv:2506.00228 (cross-list from cs.MA) [pdf, html, other]
-
Title: Sorrel: A simple and flexible framework for multi-agent reinforcement learningRebekah A. Gelpí, Yibing Ju, Ethan C. Jackson, Yikai Tang, Shon Verch, Claas Voelcker, William A. CunninghamSubjects: Multiagent Systems (cs.MA); Machine Learning (cs.LG)
We introduce Sorrel (this https URL), a simple Python interface for generating and testing new multi-agent reinforcement learning environments. This interface places a high degree of emphasis on simplicity and accessibility, and uses a more psychologically intuitive structure for the basic agent-environment loop, making it a useful tool for social scientists to investigate how learning and social interaction leads to the development and change of group dynamics. In this short paper, we outline the basic design philosophy and features of Sorrel.
- [277] arXiv:2506.00238 (cross-list from cs.CV) [pdf, other]
-
Title: ZeShot-VQA: Zero-Shot Visual Question Answering Framework with Answer Mapping for Natural Disaster Damage AssessmentComments: Accepted by the 2025 IEEE International Geoscience and Remote Sensing Symposium (IGARSS 2025)Subjects: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
Natural disasters usually affect vast areas and devastate infrastructures. Performing a timely and efficient response is crucial to minimize the impact on affected communities, and data-driven approaches are the best choice. Visual question answering (VQA) models help management teams to achieve in-depth understanding of damages. However, recently published models do not possess the ability to answer open-ended questions and only select the best answer among a predefined list of answers. If we want to ask questions with new additional possible answers that do not exist in the predefined list, the model needs to be fin-tuned/retrained on a new collected and annotated dataset, which is a time-consuming procedure. In recent years, large-scale Vision-Language Models (VLMs) have earned significant attention. These models are trained on extensive datasets and demonstrate strong performance on both unimodal and multimodal vision/language downstream tasks, often without the need for fine-tuning. In this paper, we propose a VLM-based zero-shot VQA (ZeShot-VQA) method, and investigate the performance of on post-disaster FloodNet dataset. Since the proposed method takes advantage of zero-shot learning, it can be applied on new datasets without fine-tuning. In addition, ZeShot-VQA is able to process and generate answers that has been not seen during the training procedure, which demonstrates its flexibility.
- [278] arXiv:2506.00252 (cross-list from math.OC) [pdf, html, other]
-
Title: How hard is learning to cut? Trade-offs and sample complexitySubjects: Optimization and Control (math.OC); Machine Learning (cs.LG)
In the recent years, branch-and-cut algorithms have been the target of data-driven approaches designed to enhance the decision making in different phases of the algorithm such as branching, or the choice of cutting planes (cuts). In particular, for cutting plane selection two score functions have been proposed in the literature to evaluate the quality of a cut: branch-and-cut tree size and gap closed. In this paper, we present new sample complexity lower bounds, valid for both scores. We show that for a wide family of classes $\mathcal{F}$ that maps an instance to a cut, learning over an unknown distribution of the instances to minimize those scores requires at least (up to multiplicative constants) as many samples as learning from the same class function $\mathcal{F}$ any generic target function (using square loss). Our results also extend to the case of learning from a restricted set of cuts, namely those from the Simplex tableau. To the best of our knowledge, these constitute the first lower bounds for the learning-to-cut framework. We compare our bounds to known upper bounds in the case of neural networks and show they are nearly tight. We illustrate our results with a graph neural network selection evaluated on set covering and facility location integer programming models and we empirically show that the gap closed score is an effective proxy to minimize the branch-and-cut tree size. Although the gap closed score has been extensively used in the integer programming literature, this is the first principled analysis discussing both scores at the same time both theoretically and computationally.
- [279] arXiv:2506.00270 (cross-list from stat.ML) [pdf, html, other]
-
Title: Bayesian Data Sketching for Varying Coefficient Regression ModelsSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
Varying coefficient models are popular for estimating nonlinear regression functions in functional data models. Their Bayesian variants have received limited attention in large data applications, primarily due to prohibitively slow posterior computations using Markov chain Monte Carlo (MCMC) algorithms. We introduce Bayesian data sketching for varying coefficient models to obviate computational challenges presented by large sample sizes. To address the challenges of analyzing large data, we compress the functional response vector and predictor matrix by a random linear transformation to achieve dimension reduction and conduct inference on the compressed data. Our approach distinguishes itself from several existing methods for analyzing large functional data in that it requires neither the development of new models or algorithms, nor any specialized computational hardware while delivering fully model-based Bayesian inference. Well-established methods and algorithms for varying coefficient regression models can be applied to the compressed data.
- [280] arXiv:2506.00273 (cross-list from eess.AS) [pdf, html, other]
-
Title: SoundSculpt: Direction and Semantics Driven Ambisonic Target Sound ExtractionSubjects: Audio and Speech Processing (eess.AS); Machine Learning (cs.LG); Sound (cs.SD)
This paper introduces SoundSculpt, a neural network designed to extract target sound fields from ambisonic recordings. SoundSculpt employs an ambisonic-in-ambisonic-out architecture and is conditioned on both spatial information (e.g., target direction obtained by pointing at an immersive video) and semantic embeddings (e.g., derived from image segmentation and captioning). Trained and evaluated on synthetic and real ambisonic mixtures, SoundSculpt demonstrates superior performance compared to various signal processing baselines. Our results further reveal that while spatial conditioning alone can be effective, the combination of spatial and semantic information is beneficial in scenarios where there are secondary sound sources spatially close to the target. Additionally, we compare two different semantic embeddings derived from a text description of the target sound using text encoders.
- [281] arXiv:2506.00279 (cross-list from cs.AI) [pdf, html, other]
-
Title: Sleep Brain and Cardiac Activity Predict Cognitive Flexibility and Conceptual Reasoning Using Deep LearningComments: This work was accepted for publication in IEEE EMBC 2025Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Despite extensive research on the relationship between sleep and cognition, the connection between sleep microstructure and human performance across specific cognitive domains remains underexplored. This study investigates whether deep learning models can predict executive functions, particularly cognitive adaptability and conceptual reasoning from physiological processes during a night's sleep. To address this, we introduce CogPSGFormer, a multi-scale convolutional-transformer model designed to process multi-modal polysomnographic data. This model integrates one-channel ECG and EEG signals along with extracted features, including EEG power bands and heart rate variability parameters, to capture complementary information across modalities. A thorough evaluation of the CogPSGFormer architecture was conducted to optimize the processing of extended sleep signals and identify the most effective configuration. The proposed framework was evaluated on 817 individuals from the STAGES dataset using cross-validation. The model achieved 80.3\% accuracy in classifying individuals into low vs. high cognitive performance groups on unseen data based on Penn Conditional Exclusion Test (PCET) scores. These findings highlight the effectiveness of our multi-scale feature extraction and multi-modal learning approach in leveraging sleep-derived signals for cognitive performance prediction. To facilitate reproducibility, our code is publicly accessible (this https URL).
- [282] arXiv:2506.00280 (cross-list from cs.CR) [pdf, html, other]
-
Title: 3D Gaussian Splat VulnerabilitiesMatthew Hull, Haoyang Yang, Pratham Mehta, Mansi Phute, Aeree Cho, Haoran Wang, Matthew Lau, Wenke Lee, Willian T. Lunardi, Martin Andreoni, Polo ChauComments: 4 pages, 4 figures, CVPR '25 Workshop on Neural Fields Beyond Conventional CamerasSubjects: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
With 3D Gaussian Splatting (3DGS) being increasingly used in safety-critical applications, how can an adversary manipulate the scene to cause harm? We introduce CLOAK, the first attack that leverages view-dependent Gaussian appearances - colors and textures that change with viewing angle - to embed adversarial content visible only from specific viewpoints. We further demonstrate DAGGER, a targeted adversarial attack directly perturbing 3D Gaussians without access to underlying training data, deceiving multi-stage object detectors e.g., Faster R-CNN, through established methods such as projected gradient descent. These attacks highlight underexplored vulnerabilities in 3DGS, introducing a new potential threat to robotic learning for autonomous navigation and other safety-critical 3DGS applications.
- [283] arXiv:2506.00290 (cross-list from cs.CL) [pdf, html, other]
-
Title: DLM-One: Diffusion Language Models for One-Step Sequence GenerationSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG); Machine Learning (stat.ML)
This paper introduces DLM-One, a score-distillation-based framework for one-step sequence generation with continuous diffusion language models (DLMs). DLM-One eliminates the need for iterative refinement by aligning the scores of a student model's outputs in the continuous token embedding space with the score function of a pretrained teacher DLM. We investigate whether DLM-One can achieve substantial gains in sampling efficiency for language modeling. Through comprehensive experiments on DiffuSeq -- a representative continuous DLM -- we show that DLM-One achieves up to ~500x speedup in inference time while maintaining competitive performance on benchmark text generation tasks used to evaluate the teacher models. We further analyze the method's empirical behavior across multiple datasets, providing initial insights into its generality and practical applicability. Our findings position one-step diffusion as a promising direction for efficient, high-quality language generation and broader adoption of continuous diffusion models operating in embedding space for natural language processing.
- [284] arXiv:2506.00305 (cross-list from cs.RO) [pdf, html, other]
-
Title: Learning Aerodynamics for the Control of Flying Humanoid RobotsAntonello Paolino, Gabriele Nava, Fabio Di Natale, Fabio Bergonti, Punith Reddy Vanteddu, Donato Grassi, Luca Riccobene, Alex Zanotti, Renato Tognaccini, Gianluca Iaccarino, Daniele PucciSubjects: Robotics (cs.RO); Machine Learning (cs.LG)
Robots with multi-modal locomotion are an active research field due to their versatility in diverse environments. In this context, additional actuation can provide humanoid robots with aerial capabilities. Flying humanoid robots face challenges in modeling and control, particularly with aerodynamic forces. This paper addresses these challenges from a technological and scientific standpoint. The technological contribution includes the mechanical design of iRonCub-Mk1, a jet-powered humanoid robot, optimized for jet engine integration, and hardware modifications for wind tunnel experiments on humanoid robots for precise aerodynamic forces and surface pressure measurements. The scientific contribution offers a comprehensive approach to model and control aerodynamic forces using classical and learning techniques. Computational Fluid Dynamics (CFD) simulations calculate aerodynamic forces, validated through wind tunnel experiments on iRonCub-Mk1. An automated CFD framework expands the aerodynamic dataset, enabling the training of a Deep Neural Network and a linear regression model. These models are integrated into a simulator for designing aerodynamic-aware controllers, validated through flight simulations and balancing experiments on the iRonCub-Mk1 physical prototype.
- [285] arXiv:2506.00307 (cross-list from cs.CL) [pdf, html, other]
-
Title: Lossless Token Sequence Compression via Meta-TokensComments: 16 pages, 8 figuresSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Existing work on prompt compression for Large Language Models (LLM) focuses on lossy methods that try to maximize the retention of semantic information that is relevant to downstream tasks while significantly reducing the sequence length. In this paper, we introduce a task-agnostic lossless compression technique similar to LZ77 that makes it possible to reduce the input token sequence length on average by 27\% and 18\% for the two evaluation tasks explored here. Given that we use transformer-based LLMs, this equates to 47\% and 33\% less encoding computation, respectively, due to the quadratic nature of attention. The token sequence transformation is trivial to reverse and highlights that no semantic information is lost in the process. We evaluate our proposed approach on two tasks that require strict preservation of semantics/syntax and demonstrate that existing lossy compression methods perform poorly in this setting. We find that our lossless compression technique produces only a small gap in performance compared to using the uncompressed input and posit that larger models and an expanded computing budget would likely erase the gap entirely.
- [286] arXiv:2506.00315 (cross-list from eess.SP) [pdf, html, other]
-
Title: Power-of-Two (PoT) Weights in Large Language Models (LLMs)Subjects: Signal Processing (eess.SP); Machine Learning (cs.LG)
Complexity of Neural Networks is increasing rapidly due to the massive increase in model parameters. Specifically, in Large Language Models (LLMs), the number of model parameters has grown exponentially in the past few years, for example, from 1.5 billion parameters in GPT2 to 175 billion in GPT3. This raises a significant challenge for implementation, especially for Edge devices where memory and processing power are very limited. In this work, we investigate reducing LLM complexity with special type of quantization, power of two (PoT), for linear layers weights and transformer tables. PoT not only provides memory reduction but more importantly provides significant computational reduction through converting multiplication to bit shifting. We obtained preliminary results of PoT quantization on Nano-GPT implementation using Shakespeare dataset. We then extended results to 124-M GPT-2 model. The PoT quantization results are shown to be very promising with cross entropy loss degradation $\approx$[1.3-0.88] with number of bits range [4-6] to represent power levels.
- [287] arXiv:2506.00320 (cross-list from cs.AI) [pdf, html, other]
-
Title: Dyna-Think: Synergizing Reasoning, Acting, and World Model Simulation in AI AgentsSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Recent progress in reasoning with large language models (LLMs), such as DeepSeek-R1, demonstrates impressive capabilities in domains like mathematics and coding, by exhibiting complex cognitive behaviors such as verification, goal decomposition, and self-reflection. However, it is unclear what behavior is effective and what behavior is missing for long-horizon AI agents tasks. In this work, we propose Dyna-Think, a thinking framework that integrates planning with an internal world model with reasoning and acting to enhance AI agent performance. To enable Dyna-Think, we propose Dyna-Think Imitation Learning (DIT) and Dyna-Think Dyna Training (DDT). To initialize a policy with Dyna-Think, DIT reconstructs the thinking process of R1 to focus on performing world model simulation relevant to the proposed (and planned) action, and trains the policy using this reconstructed data. To enhance Dyna-Think, DDT uses a two-stage training process to first improve the agent's world modeling ability via objectives such as state prediction or critique generation, and then improve the agent's action via policy training. We evaluate our methods on OSWorld, and demonstrate that Dyna-Think improves the agent's in-domain and out-of-domain performance, achieving similar best-of-n performance compared to R1 while generating 2x less tokens on average. Our extensive empirical studies reveal that 1) using critique generation for world model training is effective to improve policy performance; and 2) AI agents with better performance correlate with better world modeling abilities. We believe our results suggest a promising research direction to integrate world model simulation into AI agents to enhance their reasoning, planning, and acting capabilities.
- [288] arXiv:2506.00322 (cross-list from cs.CR) [pdf, html, other]
-
Title: dpmm: Differentially Private Marginal Models, a Library for Synthetic Tabular Data GenerationComments: Accepted to the Theory and Practice of Differential Privacy Workshop (TPDP 2025)Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
We propose dpmm, an open-source library for synthetic data generation with Differentially Private (DP) guarantees. It includes three popular marginal models -- PrivBayes, MST, and AIM -- that achieve superior utility and offer richer functionality compared to alternative implementations. Additionally, we adopt best practices to provide end-to-end DP guarantees and address well-known DP-related vulnerabilities. Our goal is to accommodate a wide audience with easy-to-install, highly customizable, and robust model implementations.
Our codebase is available from this https URL. - [289] arXiv:2506.00343 (cross-list from cs.SD) [pdf, html, other]
-
Title: The iNaturalist Sounds DatasetSubjects: Sound (cs.SD); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
We present the iNaturalist Sounds Dataset (iNatSounds), a collection of 230,000 audio files capturing sounds from over 5,500 species, contributed by more than 27,000 recordists worldwide. The dataset encompasses sounds from birds, mammals, insects, reptiles, and amphibians, with audio and species labels derived from observations submitted to iNaturalist, a global citizen science platform. Each recording in the dataset varies in length and includes a single species annotation. We benchmark multiple backbone architectures, comparing multiclass classification objectives with multilabel objectives. Despite weak labeling, we demonstrate that iNatSounds serves as a useful pretraining resource by benchmarking it on strongly labeled downstream evaluation datasets. The dataset is available as a single, freely accessible archive, promoting accessibility and research in this important domain. We envision models trained on this data powering next-generation public engagement applications, and assisting biologists, ecologists, and land use managers in processing large audio collections, thereby contributing to the understanding of species compositions in diverse soundscapes.
- [290] arXiv:2506.00348 (cross-list from stat.ML) [pdf, html, other]
-
Title: Beyond Winning: Margin of Victory Relative to Expectation Unlocks Accurate Skill RatingsSubjects: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Knowledge of accurate relative skills in any competitive system is essential, but foundational approaches such as ELO discard extremely relevant performance data by concentrating exclusively on binary outcomes. While margin of victory (MOV) extensions exist, they often lack a definitive method for incorporating this information. We introduce Margin of Victory Differential Analysis (MOVDA), a framework that enhances traditional rating systems by using the deviation between the true MOV and a $\textit{modeled expectation}$. MOVDA learns a domain-specific, non-linear function (a scaled hyperbolic tangent that captures saturation effects and home advantage) to predict expected MOV based on rating differentials. Crucially, the $\textit{difference}$ between the true and expected MOV provides a subtle and weighted signal for rating updates, highlighting informative deviations in all levels of contests. Extensive experiments on professional NBA basketball data (from 2013 to 2023, with 13,619 games) show that MOVDA significantly outperforms standard ELO and Bayesian baselines. MOVDA reduces Brier score prediction error by $1.54\%$ compared to TrueSkill, increases outcome accuracy by $0.58\%$, and most importantly accelerates rating convergence by $13.5\%$, while maintaining the computational efficiency of the original ELO updates. MOVDA offers a theoretically motivated, empirically superior, and computationally lean approach to integrating performance magnitude into skill rating for competitive environments like the NBA.
- [291] arXiv:2506.00358 (cross-list from cs.SD) [pdf, html, other]
-
Title: $\texttt{AVROBUSTBENCH}$: Benchmarking the Robustness of Audio-Visual Recognition Models at Test-TimeSarthak Kumar Maharana, Saksham Singh Kushwaha, Baoming Zhang, Adrian Rodriguez, Songtao Wei, Yapeng Tian, Yunhui GuoComments: Under review. For uniformity, all TTA experiments are done with a batch size of 16Subjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
While recent audio-visual models have demonstrated impressive performance, their robustness to distributional shifts at test-time remains not fully understood. Existing robustness benchmarks mainly focus on single modalities, making them insufficient for thoroughly assessing the robustness of audio-visual models. Motivated by real-world scenarios where shifts can occur $\textit{simultaneously}$ in both audio and visual modalities, we introduce $\texttt{AVROBUSTBENCH}$, a comprehensive benchmark designed to evaluate the test-time robustness of audio-visual recognition models. $\texttt{AVROBUSTBENCH}$ comprises four audio-visual benchmark datasets, $\texttt{AUDIOSET-2C}$, $\texttt{VGGSOUND-2C}$, $\texttt{KINETICS-2C}$, and $\texttt{EPICKITCHENS-2C}$, each incorporating 75 bimodal audio-visual corruptions that are $\textit{co-occurring}$ and $\textit{correlated}$. Through extensive evaluations, we observe that state-of-the-art supervised and self-supervised audio-visual models exhibit declining robustness as corruption severity increases. Furthermore, online test-time adaptation (TTA) methods, on $\texttt{VGGSOUND-2C}$ and $\texttt{KINETICS-2C}$, offer minimal improvements in performance under bimodal corruptions. We further propose $\texttt{AV2C}$, a simple TTA approach enabling on-the-fly cross-modal fusion by penalizing high-entropy samples, which achieves improvements on $\texttt{VGGSOUND-2C}$. We hope that $\texttt{AVROBUSTBENCH}$ will steer the development of more effective and robust audio-visual TTA approaches. Our code is available $\href{this https URL}{here}$.
- [292] arXiv:2506.00379 (cross-list from stat.ML) [pdf, html, other]
-
Title: Label-shift robust federated feature screening for high-dimensional classificationComments: 57 pages,9 tables,8 figuresSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
Distributed and federated learning are important tools for high-dimensional classification of large datasets. To reduce computational costs and overcome the curse of dimensionality, feature screening plays a pivotal role in eliminating irrelevant features during data preprocessing. However, data heterogeneity, particularly label shifting across different clients, presents significant challenges for feature screening. This paper introduces a general framework that unifies existing screening methods and proposes a novel utility, label-shift robust federated feature screening (LR-FFS), along with its federated estimation procedure. The framework facilitates a uniform analysis of methods and systematically characterizes their behaviors under label shift conditions. Building upon this framework, LR-FFS leverages conditional distribution functions and expectations to address label shift without adding computational burdens and remains robust against model misspecification and outliers. Additionally, the federated procedure ensures computational efficiency and privacy protection while maintaining screening effectiveness comparable to centralized processing. We also provide a false discovery rate (FDR) control method for federated feature screening. Experimental results and theoretical analyses demonstrate LR-FFS's superior performance across diverse client environments, including those with varying class distributions, sample sizes, and missing categorical data.
- [293] arXiv:2506.00385 (cross-list from cs.SD) [pdf, html, other]
-
Title: MagiCodec: Simple Masked Gaussian-Injected Codec for High-Fidelity Reconstruction and GenerationYakun Song, Jiawei Chen, Xiaobin Zhuang, Chenpeng Du, Ziyang Ma, Jian Wu, Jian Cong, Dongya Jia, Zhuo Chen, Yuping Wang, Yuxuan Wang, Xie ChenComments: 18 pages, 3 figures. The code and pre-trained models are available at this https URLSubjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
Neural audio codecs have made significant strides in efficiently mapping raw audio waveforms into discrete token representations, which are foundational for contemporary audio generative models. However, most existing codecs are optimized primarily for reconstruction quality, often at the expense of the downstream modelability of the encoded tokens. Motivated by the need to overcome this bottleneck, we introduce $\textbf{MagiCodec}$, a novel single-layer, streaming Transformer-based audio codec. MagiCodec is designed with a multistage training pipeline that incorporates Gaussian noise injection and latent regularization, explicitly targeting the enhancement of semantic expressiveness in the generated codes while preserving high reconstruction fidelity. We analytically derive the effect of noise injection in the frequency domain, demonstrating its efficacy in attenuating high-frequency components and fostering robust tokenization. Extensive experimental evaluations show that MagiCodec surpasses state-of-the-art codecs in both reconstruction quality and downstream tasks. Notably, the tokens produced by MagiCodec exhibit Zipf-like distributions, as observed in natural languages, thereby improving compatibility with language-model-based generative architectures. The code and pre-trained models are available at this https URL.
- [294] arXiv:2506.00413 (cross-list from cs.CL) [pdf, html, other]
-
Title: Accelerating Diffusion LLMs via Adaptive Parallel DecodingComments: 10 pages, 5 figuresSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Performance (cs.PF)
The generation speed of LLMs are bottlenecked by autoregressive decoding, where tokens are predicted sequentially one by one. Alternatively, diffusion large language models (dLLMs) theoretically allow for parallel token generation, but in practice struggle to achieve the speed of autoregressive models without significantly sacrificing quality. We therefore introduce adaptive parallel decoding (APD), a novel method that dynamically adjusts the number of tokens sampled in parallel. We achieve this by defining a multiplicative mixture between the dLLM marginal probabilities and the joint probability of sequences under a small auxiliary autoregressive model. This inverts the standard setup of speculative decoding, where the goal is to sample from a large autoregressive verifier by drafting from a smaller model. We further optimize APD by enabling KV caching and limiting the size of the masked input. Altogether, our method puts forward three tunable parameters to flexibly tradeoff throughput and quality. We show that APD provides markedly higher throughput with minimal quality degradations on downstream benchmarks.
- [295] arXiv:2506.00433 (cross-list from cs.CV) [pdf, html, other]
-
Title: Latent Wavelet Diffusion: Enabling 4K Image Synthesis for FreeSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
High-resolution image synthesis remains a core challenge in generative modeling, particularly in balancing computational efficiency with the preservation of fine-grained visual detail. We present Latent Wavelet Diffusion (LWD), a lightweight framework that enables any latent diffusion model to scale to ultra-high-resolution image generation (2K to 4K) for free. LWD introduces three key components: (1) a scale-consistent variational autoencoder objective that enhances the spectral fidelity of latent representations; (2) wavelet energy maps that identify and localize detail-rich spatial regions within the latent space; and (3) a time-dependent masking strategy that focuses denoising supervision on high-frequency components during training. LWD requires no architectural modifications and incurs no additional computational overhead. Despite its simplicity, it consistently improves perceptual quality and reduces FID in ultra-high-resolution image synthesis, outperforming strong baseline models. These results highlight the effectiveness of frequency-aware, signal-driven supervision as a principled and efficient approach for high-resolution generative modeling.
- [296] arXiv:2506.00446 (cross-list from stat.ML) [pdf, html, other]
-
Title: Off-Policy Evaluation of Ranking Policies via Embedding-Space User Behavior ModelingSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
Off-policy evaluation (OPE) in ranking settings with large ranking action spaces, which stems from an increase in both the number of unique actions and length of the ranking, is essential for assessing new recommender policies using only logged bandit data from previous versions. To address the high variance issues associated with existing estimators, we introduce two new assumptions: no direct effect on rankings and user behavior model on ranking embedding spaces. We then propose the generalized marginalized inverse propensity score (GMIPS) estimator with statistically desirable properties compared to existing ones. Finally, we demonstrate that the GMIPS achieves the lowest MSE. Notably, among GMIPS variants, the marginalized reward interaction IPS (MRIPS) incorporates a doubly marginalized importance weight based on a cascade behavior assumption on ranking embeddings. MRIPS effectively balances the trade-off between bias and variance, even as the ranking action spaces increase and the above assumptions may not hold, as evidenced by our experiments.
- [297] arXiv:2506.00450 (cross-list from cs.IR) [pdf, html, other]
-
Title: DV365: Extremely Long User History Modeling at InstagramWenhan Lyu, Devashish Tyagi, Yihang Yang, Ziwei Li, Ajay Somani, Karthikeyan Shanmugasundaram, Nikola Andrejevic, Ferdi Adeputra, Curtis Zeng, Arun K. Singh, Maxime Ransan, Sagar JainComments: SIGKDD 2025 acceptedSubjects: Information Retrieval (cs.IR); Machine Learning (cs.LG)
Long user history is highly valuable signal for recommendation systems, but effectively incorporating it often comes with high cost in terms of data center power consumption and GPU. In this work, we chose offline embedding over end-to-end sequence length optimization methods to enable extremely long user sequence modeling as a cost-effective solution, and propose a new user embedding learning strategy, multi-slicing and summarization, that generates highly generalizable user representation of user's long-term stable interest. History length we encoded in this embedding is up to 70,000 and on average 40,000. This embedding, named as DV365, is proven highly incremental on top of advanced attentive user sequence models deployed in Instagram. Produced by a single upstream foundational model, it is launched in 15 different models across Instagram and Threads with significant impact, and has been production battle-proven for >1 year since our first launch.
- [298] arXiv:2506.00455 (cross-list from cs.RO) [pdf, html, other]
-
Title: Diffusion Models for Increasing Accuracy in Olfaction Sensors and DatasetsSubjects: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Robotic odour source localization (OSL) is a critical capability for autonomous systems operating in complex environments. However, current OSL methods often suffer from ambiguities, particularly when robots misattribute odours to incorrect objects due to limitations in olfactory datasets and sensor resolutions. To address this challenge, we introduce a novel machine learning method using diffusion-based molecular generation to enhance odour localization accuracy that can be used by itself or with automated olfactory dataset construction pipelines with vision-language models (VLMs) This generative process of our diffusion model expands the chemical space beyond the limitations of both current olfactory datasets and the training data of VLMs, enabling the identification of potential odourant molecules not previously documented. The generated molecules can then be more accurately validated using advanced olfactory sensors which emulate human olfactory recognition through electronic sensor arrays. By integrating visual analysis, language processing, and molecular generation, our framework enhances the ability of olfaction-vision models on robots to accurately associate odours with their correct sources, thereby improving navigation and decision-making in environments where olfactory cues are essential. Our methodology represents a foundational advancement in the field of robotic olfaction, offering a scalable solution to the challenges posed by limited olfactory data and sensor ambiguities.
- [299] arXiv:2506.00462 (cross-list from cs.SD) [pdf, html, other]
-
Title: XMAD-Bench: Cross-Domain Multilingual Audio Deepfake BenchmarkIoan-Paul Ciobanu, Andrei-Iulian Hiji, Nicolae-Catalin Ristea, Paul Irofti, Cristian Rusu, Radu Tudor IonescuSubjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
Recent advances in audio generation led to an increasing number of deepfakes, making the general public more vulnerable to financial scams, identity theft, and misinformation. Audio deepfake detectors promise to alleviate this issue, with many recent studies reporting accuracy rates close to 99%. However, these methods are typically tested in an in-domain setup, where the deepfake samples from the training and test sets are produced by the same generative models. To this end, we introduce XMAD-Bench, a large-scale cross-domain multilingual audio deepfake benchmark comprising 668.8 hours of real and deepfake speech. In our novel dataset, the speakers, the generative methods, and the real audio sources are distinct across training and test splits. This leads to a challenging cross-domain evaluation setup, where audio deepfake detectors can be tested ``in the wild''. Our in-domain and cross-domain experiments indicate a clear disparity between the in-domain performance of deepfake detectors, which is usually as high as 100%, and the cross-domain performance of the same models, which is sometimes similar to random chance. Our benchmark highlights the need for the development of robust audio deepfake detectors, which maintain their generalization capacity across different languages, speakers, generative methods, and data sources. Our benchmark is publicly released at this https URL.
- [300] arXiv:2506.00471 (cross-list from physics.geo-ph) [pdf, html, other]
-
Title: DiffPINN: Generative diffusion-initialized physics-informed neural networks for accelerating seismic wavefield representationSubjects: Geophysics (physics.geo-ph); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
Physics-informed neural networks (PINNs) offer a powerful framework for seismic wavefield modeling, yet they typically require time-consuming retraining when applied to different velocity models. Moreover, their training can suffer from slow convergence due to the complexity of of the wavefield solution. To address these challenges, we introduce a latent diffusion-based strategy for rapid and effective PINN initialization. First, we train multiple PINNs to represent frequency-domain scattered wavefields for various velocity models, then flatten each trained network's parameters into a one-dimensional vector, creating a comprehensive parameter dataset. Next, we employ an autoencoder to learn latent representations of these parameter vectors, capturing essential patterns across diverse PINN's parameters. We then train a conditional diffusion model to store the distribution of these latent vectors, with the corresponding velocity models serving as conditions. Once trained, this diffusion model can generate latent vectors corresponding to new velocity models, which are subsequently decoded by the autoencoder into complete PINN parameters. Experimental results indicate that our method significantly accelerates training and maintains high accuracy across in-distribution and out-of-distribution velocity scenarios.
- [301] arXiv:2506.00479 (cross-list from cs.CL) [pdf, html, other]
-
Title: EffiVLM-BENCH: A Comprehensive Benchmark for Evaluating Training-Free Acceleration in Large Vision-Language ModelsComments: ACL 2025Subjects: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Large Vision-Language Models (LVLMs) have achieved remarkable success, yet their significant computational demands hinder practical deployment. While efforts to improve LVLM efficiency are growing, existing methods lack comprehensive evaluation across diverse backbones, benchmarks, and metrics. In this work, we systematically evaluate mainstream acceleration techniques for LVLMs, categorized into token and parameter compression. We introduce EffiVLM-Bench, a unified framework for assessing not only absolute performance but also generalization and loyalty, while exploring Pareto-optimal trade-offs. Our extensive experiments and in-depth analyses offer insights into optimal strategies for accelerating LVLMs. We open-source code and recipes for EffiVLM-Bench to foster future research.
- [302] arXiv:2506.00483 (cross-list from cs.CL) [pdf, html, other]
-
Title: Auto-Patching: Enhancing Multi-Hop Reasoning in Language ModelsComments: 8 pages, 5 figuresSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Multi-hop questions still stump large language models (LLMs), which struggle to link information across multiple reasoning steps. We introduce Auto-Patch, a novel method that dynamically patches hidden states during inference to enhance multi-hop reasoning in LLMs. Building on the PatchScopes framework, Auto-Patch selectively modifies internal representations using a learned classifier. Evaluated on the MuSiQue dataset, Auto-Patch improves the solve rate from 18.45\% (baseline) to 23.63~$\pm$~0.7\% (3 runs), narrowing the gap to Chain-of-Thought prompting (27.44\%). Our results highlight the potential of dynamic hidden state interventions for advancing complex reasoning in LLMs.
- [303] arXiv:2506.00548 (cross-list from cs.CR) [pdf, other]
-
Title: Con Instruction: Universal Jailbreaking of Multimodal Large Language Models via Non-Textual ModalitiesSubjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL); Machine Learning (cs.LG)
Existing attacks against multimodal language models (MLLMs) primarily communicate instructions through text accompanied by adversarial images. In contrast, we exploit the capabilities of MLLMs to interpret non-textual instructions, specifically, adversarial images or audio generated by our novel method, Con Instruction. We optimize these adversarial examples to align closely with target instructions in the embedding space, revealing the detrimental implications of MLLMs' sophisticated understanding. Unlike prior work, our method does not require training data or preprocessing of textual instructions. While these non-textual adversarial examples can effectively bypass MLLM safety mechanisms, their combination with various text inputs substantially amplifies attack success. We further introduce a new Attack Response Categorization (ARC) framework, which evaluates both the quality of the model's response and its relevance to the malicious instructions. Experimental results demonstrate that Con Instruction effectively bypasses safety mechanisms in multiple vision- and audio-language models, including LLaVA-v1.5, InternVL, Qwen-VL, and Qwen-Audio, evaluated on two standard benchmarks: AdvBench and SafeBench. Specifically, our method achieves the highest attack success rates, reaching 81.3% and 86.6% on LLaVA-v1.5 (13B). On the defense side, we explore various countermeasures against our attacks and uncover a substantial performance gap among existing techniques. Our implementation is made publicly available.
- [304] arXiv:2506.00557 (cross-list from stat.ML) [pdf, other]
-
Title: Score Matching With Missing DataComments: Accepted for ICML 2025 Conference Proceedings (Spotlight)Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
Score matching is a vital tool for learning the distribution of data with applications across many areas including diffusion processes, energy based modelling, and graphical model estimation. Despite all these applications, little work explores its use when data is incomplete. We address this by adapting score matching (and its major extensions) to work with missing data in a flexible setting where data can be partially missing over any subset of the coordinates. We provide two separate score matching variations for general use, an importance weighting (IW) approach, and a variational approach. We provide finite sample bounds for our IW approach in finite domain settings and show it to have especially strong performance in small sample lower dimensional cases. Complementing this, we show our variational approach to be strongest in more complex high-dimensional settings which we demonstrate on graphical model estimation tasks on both real and simulated data.
- [305] arXiv:2506.00589 (cross-list from cs.RO) [pdf, html, other]
-
Title: Constrained Stein Variational Gradient Descent for Robot Perception, Planning, and IdentificationSubjects: Robotics (cs.RO); Machine Learning (cs.LG)
Many core problems in robotics can be framed as constrained optimization problems. Often on these problems, the robotic system has uncertainty, or it would be advantageous to identify multiple high quality feasible solutions. To enable this, we present two novel frameworks for applying principles of constrained optimization to the new variational inference algorithm Stein variational gradient descent. Our general framework supports multiple types of constrained optimizers and can handle arbitrary constraints. We demonstrate on a variety of problems that we are able to learn to approximate distributions without violating constraints. Specifically, we show that we can build distributions of: robot motion plans that exactly avoid collisions, robot arm joint angles on the SE(3) manifold with exact table placement constraints, and object poses from point clouds with table placement constraints.
- [306] arXiv:2506.00659 (cross-list from cs.CR) [pdf, html, other]
-
Title: PackHero: A Scalable Graph-based Approach for Efficient Packer IdentificationSubjects: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
Anti-analysis techniques, particularly packing, challenge malware analysts, making packer identification fundamental. Existing packer identifiers have significant limitations: signature-based methods lack flexibility and struggle against dynamic evasion, while Machine Learning approaches require extensive training data, limiting scalability and adaptability. Consequently, achieving accurate and adaptable packer identification remains an open problem. This paper presents PackHero, a scalable and efficient methodology for identifying packers using a novel static approach. PackHero employs a Graph Matching Network and clustering to match and group Call Graphs from programs packed with known packers. We evaluate our approach on a public dataset of malware and benign samples packed with various packers, demonstrating its effectiveness and scalability across varying sample sizes. PackHero achieves a macro-average F1-score of 93.7% with just 10 samples per packer, improving to 98.3% with 100 samples. Notably, PackHero requires fewer samples to achieve stable performance compared to other Machine Learning-based tools. Overall, PackHero matches the performance of State-of-the-art signature-based tools, outperforming them in handling Virtualization-based packers such as Themida/Winlicense, with a recall of 100%.
- [307] arXiv:2506.00662 (cross-list from q-bio.GN) [pdf, other]
-
Title: Uncertainty-Aware Genomic Classification of Alzheimer's Disease: A Transformer-Based Ensemble Approach with Monte Carlo DropoutSubjects: Genomics (q-bio.GN); Machine Learning (cs.LG)
INTRODUCTION: Alzheimer's disease (AD) is genetically complex, complicating robust classification from genomic data. METHODS: We developed a transformer-based ensemble model (TrUE-Net) using Monte Carlo Dropout for uncertainty estimation in AD classification from whole-genome sequencing (WGS). We combined a transformer that preserves single-nucleotide polymorphism (SNP) sequence structure with a concurrent random forest using flattened genotypes. An uncertainty threshold separated samples into an uncertain (high-variance) group and a more certain (low-variance) group. RESULTS: We analyzed 1050 individuals, holding out half for testing. Overall accuracy and area under the receiver operating characteristic (ROC) curve (AUC) were 0.6514 and 0.6636, respectively. Excluding the uncertain group improved accuracy from 0.6263 to 0.7287 (10.24% increase) and F1 from 0.5843 to 0.8205 (23.62% increase). DISCUSSION: Monte Carlo Dropout-driven uncertainty helps identify ambiguous cases that may require further clinical evaluation, thus improving reliability in AD genomic classification.
- [308] arXiv:2506.00664 (cross-list from cs.AI) [pdf, html, other]
-
Title: OntoRAG: Enhancing Question-Answering through Automated Ontology Derivation from Unstructured Knowledge BasesSubjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Ontologies are pivotal for structuring knowledge bases to enhance question answering (QA) systems powered by Large Language Models (LLMs). However, traditional ontology creation relies on manual efforts by domain experts, a process that is time intensive, error prone, and impractical for large, dynamic knowledge domains. This paper introduces OntoRAG, an automated pipeline designed to derive ontologies from unstructured knowledge bases, with a focus on electrical relay documents. OntoRAG integrates advanced techniques, including web scraping, PDF parsing, hybrid chunking, information extraction, knowledge graph construction, and ontology creation, to transform unstructured data into a queryable ontology. By leveraging LLMs and graph based methods, OntoRAG enhances global sensemaking capabilities, outperforming conventional Retrieval Augmented Generation (RAG) and GraphRAG approaches in comprehensiveness and diversity. Experimental results demonstrate OntoRAGs effectiveness, achieving a comprehensiveness win rate of 85% against vector RAG and 75% against GraphRAGs best configuration. This work addresses the critical challenge of automating ontology creation, advancing the vision of the semantic web.
- [309] arXiv:2506.00674 (cross-list from cs.LO) [pdf, html, other]
-
Title: Thinking Out of the Box: Hybrid SAT Solving by Unconstrained Continuous OptimizationSubjects: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Optimization and Control (math.OC)
The Boolean satisfiability (SAT) problem lies at the core of many applications in combinatorial optimization, software verification, cryptography, and machine learning. While state-of-the-art solvers have demonstrated high efficiency in handling conjunctive normal form (CNF) formulas, numerous applications require non-CNF (hybrid) constraints, such as XOR, cardinality, and Not-All-Equal constraints. Recent work leverages polynomial representations to represent such hybrid constraints, but it relies on box constraints that can limit the use of powerful unconstrained optimizers. In this paper, we propose unconstrained continuous optimization formulations for hybrid SAT solving by penalty terms. We provide theoretical insights into when these penalty terms are necessary and demonstrate empirically that unconstrained optimizers (e.g., Adam) can enhance SAT solving on hybrid benchmarks. Our results highlight the potential of combining continuous optimization and machine-learning-based methods for effective hybrid SAT solving.
- [310] arXiv:2506.00681 (cross-list from cs.SD) [pdf, html, other]
-
Title: Learning to Upsample and Upmix Audio in the Latent DomainSubjects: Sound (cs.SD); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
Neural audio autoencoders create compact latent representations that preserve perceptually important information, serving as the foundation for both modern audio compression systems and generation approaches like next-token prediction and latent diffusion. Despite their prevalence, most audio processing operations, such as spatial and spectral up-sampling, still inefficiently operate on raw waveforms or spectral representations rather than directly on these compressed representations. We propose a framework that performs audio processing operations entirely within an autoencoder's latent space, eliminating the need to decode to raw audio formats. Our approach dramatically simplifies training by operating solely in the latent domain, with a latent L1 reconstruction term, augmented by a single latent adversarial discriminator. This contrasts sharply with raw-audio methods that typically require complex combinations of multi-scale losses and discriminators. Through experiments in bandwidth extension and mono-to-stereo up-mixing, we demonstrate computational efficiency gains of up to 100x while maintaining quality comparable to post-processing on raw audio. This work establishes a more efficient paradigm for audio processing pipelines that already incorporate autoencoders, enabling significantly faster and more resource-efficient workflows across various audio tasks.
- [311] arXiv:2506.00694 (cross-list from cs.CL) [pdf, html, other]
-
Title: Measuring Faithfulness and Abstention: An Automated Pipeline for Evaluating LLM-Generated 3-ply Case-Based Legal ArgumentsComments: 11 pages, 7th Workshop on Automated Semantic Analysis of Information in Legal Text, 16 June 2025, Chicago, ILSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Large Language Models (LLMs) demonstrate potential in complex legal tasks like argument generation, yet their reliability remains a concern. Building upon pilot work assessing LLM generation of 3-ply legal arguments using human evaluation, this paper introduces an automated pipeline to evaluate LLM performance on this task, specifically focusing on faithfulness (absence of hallucination), factor utilization, and appropriate abstention. We define hallucination as the generation of factors not present in the input case materials and abstention as the model's ability to refrain from generating arguments when instructed and no factual basis exists. Our automated method employs an external LLM to extract factors from generated arguments and compares them against the ground-truth factors provided in the input case triples (current case and two precedent cases). We evaluated eight distinct LLMs on three tests of increasing difficulty: 1) generating a standard 3-ply argument, 2) generating an argument with swapped precedent roles, and 3) recognizing the impossibility of argument generation due to lack of shared factors and abstaining. Our findings indicate that while current LLMs achieve high accuracy (over 90%) in avoiding hallucination on viable argument generation tests (Tests 1 & 2), they often fail to utilize the full set of relevant factors present in the cases. Critically, on the abstention test (Test 3), most models failed to follow instructions to stop, instead generating spurious arguments despite the lack of common factors. This automated pipeline provides a scalable method for assessing these crucial LLM behaviors, highlighting the need for improvements in factor utilization and robust abstention capabilities before reliable deployment in legal settings. Project page: this https URL.
- [312] arXiv:2506.00698 (cross-list from cs.CV) [pdf, other]
-
Title: Concept-Centric Token Interpretation for Vector-Quantized Generative ModelsComments: 17 pages, 7 figuresSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Vector-Quantized Generative Models (VQGMs) have emerged as powerful tools for image generation. However, the key component of VQGMs -- the codebook of discrete tokens -- is still not well understood, e.g., which tokens are critical to generate an image of a certain concept? This paper introduces Concept-Oriented Token Explanation (CORTEX), a novel approach for interpreting VQGMs by identifying concept-specific token combinations. Our framework employs two methods: (1) a sample-level explanation method that analyzes token importance scores in individual images, and (2) a codebook-level explanation method that explores the entire codebook to find globally relevant tokens. Experimental results demonstrate CORTEX's efficacy in providing clear explanations of token usage in the generative process, outperforming baselines across multiple pretrained VQGMs. Besides enhancing VQGMs transparency, CORTEX is useful in applications such as targeted image editing and shortcut feature detection. Our code is available at this https URL.
- [313] arXiv:2506.00708 (cross-list from cs.AI) [pdf, html, other]
-
Title: DrKGC: Dynamic Subgraph Retrieval-Augmented LLMs for Knowledge Graph Completion across General and Biomedical DomainsSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Knowledge graph completion (KGC) aims to predict missing triples in knowledge graphs (KGs) by leveraging existing triples and textual information. Recently, generative large language models (LLMs) have been increasingly employed for graph tasks. However, current approaches typically encode graph context in textual form, which fails to fully exploit the potential of LLMs for perceiving and reasoning about graph structures. To address this limitation, we propose DrKGC (Dynamic Subgraph Retrieval-Augmented LLMs for Knowledge Graph Completion). DrKGC employs a flexible lightweight model training strategy to learn structural embeddings and logical rules within the KG. It then leverages a novel bottom-up graph retrieval method to extract a subgraph for each query guided by the learned rules. Finally, a graph convolutional network (GCN) adapter uses the retrieved subgraph to enhance the structural embeddings, which are then integrated into the prompt for effective LLM fine-tuning. Experimental results on two general domain benchmark datasets and two biomedical datasets demonstrate the superior performance of DrKGC. Furthermore, a realistic case study in the biomedical domain highlights its interpretability and practical utility.
- [314] arXiv:2506.00713 (cross-list from cs.CL) [pdf, html, other]
-
Title: From Argumentative Text to Argument Knowledge Graph: A New Framework for Structured ArgumentationComments: 16 pages, 7 figuresSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
This paper presents a framework to convert argumentative texts into argument knowledge graphs (AKG). Starting with basic annotations of argumentative components (ACs) and argumentative relations (ARs), we enrich the information by constructing a knowledge base (KB) graph with metadata attributes for nodes. Next, we use premises and inference rules from the KB to form arguments by applying modus ponens. From these arguments, we create an AKG. The nodes and edges of the AKG have attributes that capture important argumentative features. We also find missing inference rules by identifying markers. This makes it possible to identify undercut attacks that were previously undetectable in existing datasets. The AKG gives a graphical view of the argumentative structure that is easier to understand than theoretical formats. It also prepares the ground for future reasoning tasks, including checking the coherence of arguments and identifying opportunities for revision. For this, it is important to find indirect relations, many of which are implicit. Our proposed AKG format, with annotated inference rules and modus ponens, will help reasoning models learn the implicit indirect relations that require inference over arguments and the relations between them.
- [315] arXiv:2506.00721 (cross-list from cs.CV) [pdf, html, other]
-
Title: Common Inpainted Objects In-N-Out of ContextComments: 12 pages, 7 figuresSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
We present Common Inpainted Objects In-N-Out of Context (COinCO), a novel dataset addressing the scarcity of out-of-context examples in existing vision datasets. By systematically replacing objects in COCO images through diffusion-based inpainting, we create 97,722 unique images featuring both contextually coherent and inconsistent scenes, enabling effective context learning. Each inpainted object is meticulously verified and categorized as in- or out-of-context through a multimodal large language model assessment. Our analysis reveals significant patterns in semantic priors that influence inpainting success across object categories. We demonstrate three key tasks enabled by COinCO: (1) training context classifiers that effectively determine whether existing objects belong in their context; (2) a novel Objects-from-Context prediction task that determines which new objects naturally belong in given scenes at both instance and clique levels, and (3) context-enhanced fake detection on state-of-the-art methods without fine-tuning. COinCO provides a controlled testbed with contextual variations, establishing a foundation for advancing context-aware visual understanding in computer vision and image forensics. Our code and data are at: this https URL.
- [316] arXiv:2506.00725 (cross-list from cond-mat.mtrl-sci) [pdf, html, other]
-
Title: A Foundation Model for Non-Destructive Defect Identification from Vibrational SpectraMouyang Cheng, Chu-Liang Fu, Bowen Yu, Eunbi Rha, Abhijatmedhi Chotrattanapituk, Douglas L Abernathy, Yongqiang Cheng, Mingda LiComments: 14 pages, 5 figuresSubjects: Materials Science (cond-mat.mtrl-sci); Machine Learning (cs.LG)
Defects are ubiquitous in solids and strongly influence materials' mechanical and functional properties. However, non-destructive characterization and quantification of defects, especially when multiple types coexist, remain a long-standing challenge. Here we introduce DefectNet, a foundation machine learning model that predicts the chemical identity and concentration of substitutional point defects with multiple coexisting elements directly from vibrational spectra, specifically phonon density-of-states (PDoS). Trained on over 16,000 simulated spectra from 2,000 semiconductors, DefectNet employs a tailored attention mechanism to identify up to six distinct defect elements at concentrations ranging from 0.2% to 25%. The model generalizes well to unseen crystals across 56 elements and can be fine-tuned on experimental data. Validation using inelastic scattering measurements of SiGe alloys and MgB$_2$ superconductor demonstrates its accuracy and transferability. Our work establishes vibrational spectroscopy as a viable, non-destructive probe for point defect quantification in bulk materials, and highlights the promise of foundation models in data-driven defect engineering.
- [317] arXiv:2506.00751 (cross-list from cs.AI) [pdf, html, other]
-
Title: Alignment Revisited: Are Large Language Models Consistent in Stated and Revealed Preferences?Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Recent advances in Large Language Models (LLMs) highlight the need to align their behaviors with human values. A critical, yet understudied, issue is the potential divergence between an LLM's stated preferences (its reported alignment with general principles) and its revealed preferences (inferred from decisions in contextualized scenarios). Such deviations raise fundamental concerns for the interpretability, trustworthiness, reasoning transparency, and ethical deployment of LLMs, particularly in high-stakes applications. This work formally defines and proposes a method to measure this preference deviation. We investigate how LLMs may activate different guiding principles in specific contexts, leading to choices that diverge from previously stated general principles. Our approach involves crafting a rich dataset of well-designed prompts as a series of forced binary choices and presenting them to LLMs. We compare LLM responses to general principle prompts stated preference with LLM responses to contextualized prompts revealed preference, using metrics like KL divergence to quantify the deviation. We repeat the analysis across different categories of preferences and on four mainstream LLMs and find that a minor change in prompt format can often pivot the preferred choice regardless of the preference categories and LLMs in the test. This prevalent phenomenon highlights the lack of understanding and control of the LLM decision-making competence. Our study will be crucial for integrating LLMs into services, especially those that interact directly with humans, where morality, fairness, and social responsibilities are crucial dimensions. Furthermore, identifying or being aware of such deviation will be critically important as LLMs are increasingly envisioned for autonomous agentic tasks where continuous human evaluation of all LLMs' intermediary decision-making steps is impossible.
- [318] arXiv:2506.00785 (cross-list from cs.AI) [pdf, html, other]
-
Title: GeoChain: Multimodal Chain-of-Thought for Geographic ReasoningSubjects: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
This paper introduces GeoChain, a large-scale benchmark for evaluating step-by-step geographic reasoning in multimodal large language models (MLLMs). Leveraging 1.46 million Mapillary street-level images, GeoChain pairs each image with a 21-step chain-of-thought (CoT) question sequence (over 30 million Q&A pairs). These sequences guide models from coarse attributes to fine-grained localization across four reasoning categories - visual, spatial, cultural, and precise geolocation - annotated by difficulty. Images are also enriched with semantic segmentation (150 classes) and a visual locatability score. Our benchmarking of contemporary MLLMs (GPT-4.1 variants, Claude 3.7, Gemini 2.5 variants) on a diverse 2,088-image subset reveals consistent challenges: models frequently exhibit weaknesses in visual grounding, display erratic reasoning, and struggle to achieve accurate localization, especially as the reasoning complexity escalates. GeoChain offers a robust diagnostic methodology, critical for fostering significant advancements in complex geographic reasoning within MLLMs.
- [319] arXiv:2506.00800 (cross-list from eess.AS) [pdf, html, other]
-
Title: CLAP-ART: Automated Audio Captioning with Semantic-rich Audio Representation TokenizerComments: Accepted to Interspeech2025Subjects: Audio and Speech Processing (eess.AS); Machine Learning (cs.LG); Sound (cs.SD)
Automated Audio Captioning (AAC) aims to describe the semantic contexts of general sounds, including acoustic events and scenes, by leveraging effective acoustic features. To enhance performance, an AAC method, EnCLAP, employed discrete tokens from EnCodec as an effective input for fine-tuning a language model BART. However, EnCodec is designed to reconstruct waveforms rather than capture the semantic contexts of general sounds, which AAC should describe. To address this issue, we propose CLAP-ART, an AAC method that utilizes ``semantic-rich and discrete'' tokens as input. CLAP-ART computes semantic-rich discrete tokens from pre-trained audio representations through vector quantization. We experimentally confirmed that CLAP-ART outperforms baseline EnCLAP on two AAC benchmarks, indicating that semantic-rich discrete tokens derived from semantically rich AR are beneficial for AAC.
- [320] arXiv:2506.00813 (cross-list from cs.CV) [pdf, html, other]
-
Title: TIME: TabPFN-Integrated Multimodal Engine for Robust Tabular-Image LearningSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Tabular-image multimodal learning, which integrates structured tabular data with imaging data, holds great promise for a variety of tasks, especially in medical applications. Yet, two key challenges remain: (1) the lack of a standardized, pretrained representation for tabular data, as is commonly available in vision and language domains; and (2) the difficulty of handling missing values in the tabular modality, which are common in real-world medical datasets. To address these issues, we propose the TabPFN-Integrated Multimodal Engine (TIME), a novel multimodal framework that builds on the recently introduced tabular foundation model, TabPFN. TIME leverages TabPFN as a frozen tabular encoder to generate robust, strong embeddings that are naturally resilient to missing data, and combines them with image features from pretrained vision backbones. We explore a range of fusion strategies and tabular encoders, and evaluate our approach on both natural and medical datasets. Extensive experiments demonstrate that TIME consistently outperforms competitive baselines across both complete and incomplete tabular inputs, underscoring its practical value in real-world multimodal learning scenarios.
- [321] arXiv:2506.00818 (cross-list from stat.ML) [pdf, html, other]
-
Title: Generalized Linear Markov Decision ProcessComments: 34 pages, 9 figuresSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
The linear Markov Decision Process (MDP) framework offers a principled foundation for reinforcement learning (RL) with strong theoretical guarantees and sample efficiency. However, its restrictive assumption-that both transition dynamics and reward functions are linear in the same feature space-limits its applicability in real-world domains, where rewards often exhibit nonlinear or discrete structures. Motivated by applications such as healthcare and e-commerce, where data is scarce and reward signals can be binary or count-valued, we propose the Generalized Linear MDP (GLMDP) framework-an extension of the linear MDP framework-that models rewards using generalized linear models (GLMs) while maintaining linear transition dynamics. We establish the Bellman completeness of GLMDPs with respect to a new function class that accommodates nonlinear rewards and develop two offline RL algorithms: Generalized Pessimistic Value Iteration (GPEVI) and a semi-supervised variant (SS-GPEVI) that utilizes both labeled and unlabeled trajectories. Our algorithms achieve theoretical guarantees on policy suboptimality and demonstrate improved sample efficiency in settings where reward labels are expensive or limited.
- [322] arXiv:2506.00826 (cross-list from cs.CL) [pdf, html, other]
-
Title: HERGC: Heterogeneous Experts Representation and Generative Completion for Multimodal Knowledge GraphsSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Multimodal knowledge graphs (MMKGs) enrich traditional knowledge graphs (KGs) by incorporating diverse modalities such as images and text. Multi-modal knowledge graph completion (MMKGC) seeks to exploit these heterogeneous signals to infer missing facts, thereby mitigating the intrinsic incompleteness of MMKGs. Existing MMKGC methods typically leverage only the information contained in the MMKGs under the closed-world assumption and adopt discriminative training objectives, which limits their reasoning capacity during completion. Recent generative completion approaches powered by advanced large language models (LLMs) have shown strong reasoning abilities in unimodal knowledge graph completion, but their potential in MMKGC remains largely unexplored. To bridge this gap, we propose HERGC, a Heterogeneous Experts Representation and Generative Completion framework for MMKGs. HERGC first deploys a Heterogeneous Experts Representation Retriever that enriches and fuses multimodal information and retrieves a compact candidate set for each incomplete triple. It then uses a Generative LLM Predictor fine-tuned on minimal instruction data to accurately identify the correct answer from these candidates. Extensive experiments on three standard MMKG benchmarks demonstrate HERGC's effectiveness and robustness, achieving state-of-the-art performance.
- [323] arXiv:2506.00828 (cross-list from cs.IR) [pdf, html, other]
-
Title: Breaker: Removing Shortcut Cues with User Clustering for Single-slot Recommendation SystemSubjects: Information Retrieval (cs.IR); Machine Learning (cs.LG)
In a single-slot recommendation system, users are only exposed to one item at a time, and the system cannot collect user feedback on multiple items simultaneously. Therefore, only pointwise modeling solutions can be adopted, focusing solely on modeling the likelihood of clicks or conversions for items by users to learn user-item preferences, without the ability to capture the ranking information among different items directly. However, since user-side information is often much more abundant than item-side information, the model can quickly learn the differences in user intrinsic tendencies, which are independent of the items they are exposed to. This can cause these intrinsic tendencies to become a shortcut bias for the model, leading to insufficient mining of the most concerned user-item preferences. To solve this challenge, we introduce the Breaker model. Breaker integrates an auxiliary task of user representation clustering with a multi-tower structure for cluster-specific preference modeling. By clustering user representations, we ensure that users within each cluster exhibit similar characteristics, which increases the complexity of the pointwise recommendation task on the user side. This forces the multi-tower structure with cluster-driven parameter learning to better model user-item preferences, ultimately eliminating shortcut biases related to user intrinsic tendencies. In terms of training, we propose a delayed parameter update mechanism to enhance training stability and convergence, enabling end-to-end joint training of the auxiliary clustering and classification tasks. Both offline and online experiments demonstrate that our method surpasses the baselines. It has already been deployed and is actively serving tens of millions of users daily on Meituan, one of the most popular e-commerce platforms for services.
- [324] arXiv:2506.00829 (cross-list from cs.CL) [pdf, html, other]
-
Title: COMPKE: Complex Question Answering under Knowledge EditingComments: Accepted by ACL 2025 FindingsSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Knowledge Editing, which efficiently modifies the knowledge in large language models, has gathered great attention. Current benchmarks primarily use multi-hop question answering to assess and analyze newly injected or updated knowledge. However, we argue that these benchmarks fail to effectively evaluate how well the updated models apply this knowledge in real-life scenarios, particularly when questions require complex reasoning, involving one-to-many relationships or multi-step logical intersections. To fill in this gap, we introduce a new benchmark, COMPKE: Complex Question Answering under Knowledge Editing, which includes 11,924 complex questions that reflect real-life situations. We conduct an extensive evaluation of four knowledge editing methods on COMPKE, revealing that their effectiveness varies notably across different models. For instance, MeLLo attains an accuracy of 39.47 on GPT-4O-MINI, but this drops sharply to 3.83 on QWEN2.5-3B. We further investigate the underlying causes of these disparities from both methodological and model-specific perspectives. The datasets are available at this https URL.
- [325] arXiv:2506.00839 (cross-list from cs.GR) [pdf, html, other]
-
Title: Neural Path Guiding with Distribution FactorizationComments: 11 pages, 11 figures. Accepted to EGSR 2025Subjects: Graphics (cs.GR); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
In this paper, we present a neural path guiding method to aid with Monte Carlo (MC) integration in rendering. Existing neural methods utilize distribution representations that are either fast or expressive, but not both. We propose a simple, but effective, representation that is sufficiently expressive and reasonably fast. Specifically, we break down the 2D distribution over the directional domain into two 1D probability distribution functions (PDF). We propose to model each 1D PDF using a neural network that estimates the distribution at a set of discrete coordinates. The PDF at an arbitrary location can then be evaluated and sampled through interpolation. To train the network, we maximize the similarity of the learned and target distributions. To reduce the variance of the gradient during optimizations and estimate the normalization factor, we propose to cache the incoming radiance using an additional network. Through extensive experiments, we demonstrate that our approach is better than the existing methods, particularly in challenging scenes with complex light transport.
- [326] arXiv:2506.00854 (cross-list from cs.CL) [pdf, html, other]
-
Title: EEG2TEXT-CN: An Exploratory Study of Open-Vocabulary Chinese Text-EEG Alignment via Large Language Model and Contrastive Learning on ChineseEEGSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multimedia (cs.MM); Neurons and Cognition (q-bio.NC)
We propose EEG2TEXT-CN, which, to the best of our knowledge, represents one of the earliest open-vocabulary EEG-to-text generation frameworks tailored for Chinese. Built on a biologically grounded EEG encoder (NICE-EEG) and a compact pretrained language model (MiniLM), our architecture aligns multichannel brain signals with natural language representations via masked pretraining and contrastive learning. Using a subset of the ChineseEEG dataset, where each sentence contains approximately ten Chinese characters aligned with 128-channel EEG recorded at 256 Hz, we segment EEG into per-character embeddings and predict full sentences in a zero-shot setting. The decoder is trained with teacher forcing and padding masks to accommodate variable-length sequences. Evaluation on over 1,500 training-validation sentences and 300 held-out test samples shows promising lexical alignment, with a best BLEU-1 score of 6.38\%. While syntactic fluency remains a challenge, our findings demonstrate the feasibility of non-phonetic, cross-modal language decoding from EEG. This work opens a new direction in multilingual brain-to-text research and lays the foundation for future cognitive-language interfaces in Chinese.
- [327] arXiv:2506.00863 (cross-list from cs.CL) [pdf, html, other]
-
Title: L3Cube-MahaEmotions: A Marathi Emotion Recognition Dataset with Synthetic Annotations using CoTR prompting and Large Language ModelsSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Emotion recognition in low-resource languages like Marathi remains challenging due to limited annotated data. We present L3Cube-MahaEmotions, a high-quality Marathi emotion recognition dataset with 11 fine-grained emotion labels. The training data is synthetically annotated using large language models (LLMs), while the validation and test sets are manually labeled to serve as a reliable gold-standard benchmark. Building on the MahaSent dataset, we apply the Chain-of-Translation (CoTR) prompting technique, where Marathi sentences are translated into English and emotion labeled via a single prompt. GPT-4 and Llama3-405B were evaluated, with GPT-4 selected for training data annotation due to superior label quality. We evaluate model performance using standard metrics and explore label aggregation strategies (e.g., Union, Intersection). While GPT-4 predictions outperform fine-tuned BERT models, BERT-based models trained on synthetic labels fail to surpass GPT-4. This highlights both the importance of high-quality human-labeled data and the inherent complexity of emotion recognition. An important finding of this work is that generic LLMs like GPT-4 and Llama3-405B generalize better than fine-tuned BERT for complex low-resource emotion recognition tasks. The dataset and model are shared publicly at this https URL
- [328] arXiv:2506.00866 (cross-list from stat.ML) [pdf, html, other]
-
Title: Projection Pursuit Density Ratio EstimationSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
Density ratio estimation (DRE) is a paramount task in machine learning, for its broad applications across multiple domains, such as covariate shift adaptation, causal inference, independence tests and beyond. Parametric methods for estimating the density ratio possibly lead to biased results if models are misspecified, while conventional non-parametric methods suffer from the curse of dimensionality when the dimension of data is large. To address these challenges, in this paper, we propose a novel approach for DRE based on the projection pursuit (PP) approximation. The proposed method leverages PP to mitigate the impact of high dimensionality while retaining the model flexibility needed for the accuracy of DRE. We establish the consistency and the convergence rate for the proposed estimator. Experimental results demonstrate that our proposed method outperforms existing alternatives in various applications.
- [329] arXiv:2506.00894 (cross-list from cs.SE) [pdf, html, other]
-
Title: CODEMENV: Benchmarking Large Language Models on Code MigrationKeyuan Cheng, Xudong Shen, Yihao Yang, Tengyue Wang, Yang Cao, Muhammad Asif Ali, Hanbin Wang, Lijie Hu, Di WangComments: Accepted by ACL 2025 FindingsSubjects: Software Engineering (cs.SE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Large language models (LLMs) have shown remarkable capabilities across various software engineering tasks; however, their effectiveness in code migration, adapting code to run in different environments, remains insufficiently studied. In this work, we introduce CODEMENV: Code Migration Across Environment, a new benchmark specifically designed to assess LLMs' abilities in code migration scenarios. CODEMENV consists of 922 examples spanning 19 Python and Java packages, and covers three core tasks: (1) identifying functions incompatible with specific versions, (2) detecting changes in function definitions, and (3) adapting code to target environments. Experimental evaluation with seven LLMs on CODEMENV yields an average pass@1 rate of 26.50%, with GPT-4O achieving the highest score at 43.84%. Key findings include: (i) LLMs tend to be more proficient with newer function versions, which aids in migrating legacy code, and (ii) LLMs sometimes exhibit logical inconsistencies by identifying function changes irrelevant to the intended migration environment. The datasets are available at this https URL.
- [330] arXiv:2506.00904 (cross-list from cs.CV) [pdf, html, other]
-
Title: Towards Edge-Based Idle State Detection in Construction Machinery Using Surveillance CamerasComments: 18 pages, 6 figures, 3 tables; to appear in Intelligent Systems and Applications, Lecture Notes in Networks and Systems (LNNS), Springer, 2025. Part of the 11th Intelligent Systems Conference (IntelliSys 2025), 28-29 August 2025, Amsterdam, The NetherlandsSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
The construction industry faces significant challenges in optimizing equipment utilization, as underused machinery leads to increased operational costs and project delays. Accurate and timely monitoring of equipment activity is therefore key to identifying idle periods and improving overall efficiency. This paper presents the Edge-IMI framework for detecting idle construction machinery, specifically designed for integration with surveillance camera systems. The proposed solution consists of three components: object detection, tracking, and idle state identification, which are tailored for execution on resource-constrained, CPU-based edge computing devices. The performance of Edge-IMI is evaluated using a combined dataset derived from the ACID and MOCS benchmarks. Experimental results confirm that the object detector achieves an F1 score of 71.75%, indicating robust real-world detection capabilities. The logistic regression-based idle identification module reliably distinguishes between active and idle machinery with minimal false positives. Integrating all three modules, Edge-IMI enables efficient on-site inference, reducing reliance on high-bandwidth cloud services and costly hardware accelerators. We also evaluate the performance of object detection models on Raspberry Pi 5 and an Intel NUC platforms, as example edge computing platforms. We assess the feasibility of real-time processing and the impact of model optimization techniques.
- [331] arXiv:2506.00925 (cross-list from q-bio.BM) [pdf, html, other]
-
Title: ProtInvTree: Deliberate Protein Inverse Folding with Reward-guided Tree SearchSubjects: Biomolecules (q-bio.BM); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Designing protein sequences that fold into a target 3D structure, known as protein inverse folding, is a fundamental challenge in protein engineering. While recent deep learning methods have achieved impressive performance by recovering native sequences, they often overlook the one-to-many nature of the problem: multiple diverse sequences can fold into the same structure. This motivates the need for a generative model capable of designing diverse sequences while preserving structural consistency. To address this trade-off, we introduce ProtInvTree, the first reward-guided tree-search framework for protein inverse folding. ProtInvTree reformulates sequence generation as a deliberate, step-wise decision-making process, enabling the exploration of multiple design paths and exploitation of promising candidates through self-evaluation, lookahead, and backtracking. We propose a two-stage focus-and-grounding action mechanism that decouples position selection and residue generation. To efficiently evaluate intermediate states, we introduce a jumpy denoising strategy that avoids full rollouts. Built upon pretrained protein language models, ProtInvTree supports flexible test-time scaling by expanding the search depth and breadth without retraining. Empirically, ProtInvTree outperforms state-of-the-art baselines across multiple benchmarks, generating structurally consistent yet diverse sequences, including those far from the native ground truth.
- [332] arXiv:2506.00933 (cross-list from stat.ML) [pdf, html, other]
-
Title: Reconstruction and Prediction of Volterra Integral Equations Driven by Gaussian NoiseSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
Integral equations are widely used in fields such as applied modeling, medical imaging, and system identification, providing a powerful framework for solving deterministic problems. While parameter identification for differential equations has been extensively studied, the focus on integral equations, particularly stochastic Volterra integral equations, remains limited. This research addresses the parameter identification problem, also known as the equation reconstruction problem, in Volterra integral equations driven by Gaussian noise. We propose an improved deep neural networks framework for estimating unknown parameters in the drift term of these equations. The network represents the primary variables and their integrals, enhancing parameter estimation accuracy by incorporating inter-output relationships into the loss function. Additionally, the framework extends beyond parameter identification to predict the system's behavior outside the integration interval. Prediction accuracy is validated by comparing predicted and true trajectories using a 95% confidence interval. Numerical experiments demonstrate the effectiveness of the proposed deep neural networks framework in both parameter identification and prediction tasks, showing robust performance under varying noise levels and providing accurate solutions for modeling stochastic systems.
- [333] arXiv:2506.00983 (cross-list from cs.IR) [pdf, html, other]
-
Title: Bridging the Gap: From Ad-hoc to Proactive Search in ConversationsComments: Accepted as a full paper at SIGIR 2025Subjects: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Proactive search in conversations (PSC) aims to reduce user effort in formulating explicit queries by proactively retrieving useful relevant information given conversational context. Previous work in PSC either directly uses this context as input to off-the-shelf ad-hoc retrievers or further fine-tunes them on PSC data. However, ad-hoc retrievers are pre-trained on short and concise queries, while the PSC input is longer and noisier. This input mismatch between ad-hoc search and PSC limits retrieval quality. While fine-tuning on PSC data helps, its benefits remain constrained by this input gap. In this work, we propose Conv2Query, a novel conversation-to-query framework that adapts ad-hoc retrievers to PSC by bridging the input gap between ad-hoc search and PSC. Conv2Query maps conversational context into ad-hoc queries, which can either be used as input for off-the-shelf ad-hoc retrievers or for further fine-tuning on PSC data. Extensive experiments on two PSC datasets show that Conv2Query significantly improves ad-hoc retrievers' performance, both when used directly and after fine-tuning on PSC.
- [334] arXiv:2506.01034 (cross-list from cs.CL) [pdf, html, other]
-
Title: Less is More: Local Intrinsic Dimensions of Contextual Language ModelsBenjamin Matthias Ruppik, Julius von Rohrscheidt, Carel van Niekerk, Michael Heck, Renato Vukovic, Shutong Feng, Hsien-chin Lin, Nurul Lubis, Bastian Rieck, Marcus Zibrowius, Milica GašićComments: 9 pages, with an additional 13 pages of appendixSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Understanding the internal mechanisms of large language models (LLMs) remains a challenging and complex endeavor. Even fundamental questions, such as how fine-tuning affects model behavior, often require extensive empirical evaluation. In this paper, we introduce a novel perspective based on the geometric properties of contextual latent embeddings to study the effects of training and fine-tuning. To that end, we measure the local dimensions of a contextual language model's latent space and analyze their shifts during training and fine-tuning. We show that the local dimensions provide insights into the model's training dynamics and generalization ability. Specifically, the mean of the local dimensions predicts when the model's training capabilities are exhausted, as exemplified in a dialogue state tracking task, overfitting, as demonstrated in an emotion recognition task, and grokking, as illustrated with an arithmetic task. Furthermore, our experiments suggest a practical heuristic: reductions in the mean local dimension tend to accompany and predict subsequent performance gains. Through this exploration, we aim to provide practitioners with a deeper understanding of the implications of fine-tuning on embedding spaces, facilitating informed decisions when configuring models for specific applications. The results of this work contribute to the ongoing discourse on the interpretability, adaptability, and generalizability of LLMs by bridging the gap between intrinsic model mechanisms and geometric properties in the respective embeddings.
- [335] arXiv:2506.01062 (cross-list from cs.CL) [pdf, html, other]
-
Title: SealQA: Raising the Bar for Reasoning in Search-Augmented Language ModelsComments: Preprint. 22 pages, 7 figures, 11 tablesSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
We introduce SealQA, a new challenge benchmark for evaluating SEarch-Augmented Language models on fact-seeking questions where web search yields conflicting, noisy, or unhelpful results. SealQA comes in three flavors: (1) Seal-0 (main) and (2) Seal-Hard, which assess factual accuracy and reasoning capabilities, with Seal-0 focusing on the most challenging questions where chat models (e.g., GPT-4.1) typically achieve near-zero accuracy; and (3) LongSeal, which extends SealQA to test long-context, multi-document reasoning in "needle-in-a-haystack" settings. Our evaluation reveals critical limitations in current models: Even frontier LLMs perform poorly across all SealQA flavors. On Seal-0, frontier agentic models equipped with tools like o3 and o4-mini achieve only 17.1% and 6.3% accuracy, respectively, at their best reasoning efforts. We find that advanced reasoning models such as DeepSeek-R1-671B and o3-mini are highly vulnerable to noisy search results. Notably, increasing test-time compute does not yield reliable gains across o3-mini, o4-mini, and o3, with performance often plateauing or even declining early. Additionally, while recent models are less affected by the "lost-in-the-middle" issue, they still fail to reliably identify relevant documents in LongSeal when faced with numerous distractors. To facilitate future work, we release SealQA at this http URL.
- [336] arXiv:2506.01069 (cross-list from cs.CV) [pdf, other]
-
Title: Revolutionizing Blood Banks: AI-Driven Fingerprint-Blood Group Correlation for Enhanced SafetyJournal-ref: Data and Metadata [Internet]. 2025 Apr. 7 [cited 2025 Jun. 1];4:894Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Identification of a person is central in forensic science, security, and healthcare. Methods such as iris scanning and genomic profiling are more accurate but expensive, time-consuming, and more difficult to implement. This study focuses on the relationship between the fingerprint patterns and the ABO blood group as a biometric identification tool. A total of 200 subjects were included in the study, and fingerprint types (loops, whorls, and arches) and blood groups were compared. Associations were evaluated with statistical tests, including chi-square and Pearson correlation. The study found that the loops were the most common fingerprint pattern and the O+ blood group was the most prevalent. Even though there was some associative pattern, there was no statistically significant difference in the fingerprint patterns of different blood groups. Overall, the results indicate that blood group data do not significantly improve personal identification when used in conjunction with fingerprinting. Although the study shows weak correlation, it may emphasize the efforts of multi-modal based biometric systems in enhancing the current biometric systems. Future studies may focus on larger and more diverse samples, and possibly machine learning and additional biometrics to improve identification methods. This study addresses an element of the ever-changing nature of the fields of forensic science and biometric identification, highlighting the importance of resilient analytical methods for personal identification.
- [337] arXiv:2506.01075 (cross-list from cs.DS) [pdf, html, other]
-
Title: Learning DNF through Generalized Fourier RepresentationsComments: 54 pagesSubjects: Data Structures and Algorithms (cs.DS); Information Theory (cs.IT); Machine Learning (cs.LG)
The Fourier representation for the uniform distribution over the Boolean cube has found numerous applications in algorithms and complexity analysis. Notably, in learning theory, learnability of Disjunctive Normal Form (DNF) under uniform as well as product distributions has been established through such representations. This paper makes five main contributions. First, it introduces a generalized Fourier expansion that can be used with any distribution $D$ through the representation of the distribution as a Bayesian network (BN). Second, it shows that the main algorithmic tools for learning with the Fourier representation, that use membership queries to approximate functions by recovering their heavy Fourier coefficients, can be used with slight modifications with the generalized expansion. These results hold for any distribution. Third, it analyzes the $L_1$ spectral norm of conjunctions under the new expansion, showing that it is bounded for a class of distributions which can be represented by difference bounded tree BN, where a parent node in the BN representation can change the conditional expectation of a child node by at most $\alpha<0.5$. Lower bounds are presented to show that such constraints are necessary. The fourth contribution uses these results to show the learnability of DNF with membership queries under difference bounded tree BN. The final contribution is to develop an algorithm for learning difference-bounded tree BN distributions, thus extending the DNF learnability result to cases where the distribution is not known in advance.
- [338] arXiv:2506.01083 (cross-list from stat.ML) [pdf, html, other]
-
Title: Generative diffusion posterior sampling for informative likelihoodsComments: Commemorative issueSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Systems and Control (eess.SY)
Sequential Monte Carlo (SMC) methods have recently shown successful results for conditional sampling of generative diffusion models. In this paper we propose a new diffusion posterior SMC sampler achieving improved statistical efficiencies, particularly under outlier conditions or highly informative likelihoods. The key idea is to construct an observation path that correlates with the diffusion model and to design the sampler to leverage this correlation for more efficient sampling. Empirical results conclude the efficiency.
- [339] arXiv:2506.01084 (cross-list from cs.CL) [pdf, html, other]
-
Title: zip2zip: Inference-Time Adaptive Vocabularies for Language Models via Token CompressionSaibo Geng, Nathan Ranchin, Yunzhen yao, Maxime Peyrard, Chris Wendler, Michael Gastpar, Robert WestComments: Code will be released at this https URLSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Tokenization efficiency plays a critical role in the performance and cost of large language models (LLMs), yet most models rely on static tokenizers optimized for general-purpose corpora. These tokenizers' fixed vocabularies often fail to adapt to domain- or language-specific inputs, leading to longer token sequences and higher computational costs. We introduce zip2zip, a framework that enables LLMs to dynamically adjust token vocabulary at inference time, allowing for fewer generated tokens and thus faster inference. zip2zip consists of three key components: (1) a tokenizer based on Lempel-Ziv-Welch (LZW) compression that incrementally compresses tokens into reusable "hypertokens" on the fly; (2) an embedding layer that computes embeddings for newly formed hypertokens at runtime; and (3) a causal language modeling variant that trains the model to operate on hypertokenized, compressed sequences. We show that an existing LLM can be zip2zip-fied in 10 GPU-hours via parameter-efficient finetuning. The resulting zip2zip LLMs effectively learn to use hypertokens at inference time, reducing input and output sequence length by 20-60\%, with significant improvements in inference latency.
- [340] arXiv:2506.01093 (cross-list from cs.AI) [pdf, html, other]
-
Title: Regulatory Graphs and GenAI for Real-Time Transaction Monitoring and Compliance Explanation in BankingSubjects: Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG)
This paper presents a real-time transaction monitoring framework that integrates graph-based modeling, narrative field embedding, and generative explanation to support automated financial compliance. The system constructs dynamic transaction graphs, extracts structural and contextual features, and classifies suspicious behavior using a graph neural network. A retrieval-augmented generation module generates natural language explanations aligned with regulatory clauses for each flagged transaction. Experiments conducted on a simulated stream of financial data show that the proposed method achieves superior results, with 98.2% F1-score, 97.8% precision, and 97.0% recall. Expert evaluation further confirms the quality and interpretability of generated justifications. The findings demonstrate the potential of combining graph intelligence and generative models to support explainable, audit-ready compliance in high-risk financial environments.
- [341] arXiv:2506.01143 (cross-list from stat.ML) [pdf, html, other]
-
Title: Linear regression with overparameterized linear neural networks: Tight upper and lower bounds for implicit $\ell^1$-regularizationSubjects: Machine Learning (stat.ML); Information Theory (cs.IT); Machine Learning (cs.LG); Optimization and Control (math.OC)
Modern machine learning models are often trained in a setting where the number of parameters exceeds the number of training samples. To understand the implicit bias of gradient descent in such overparameterized models, prior work has studied diagonal linear neural networks in the regression setting. These studies have shown that, when initialized with small weights, gradient descent tends to favor solutions with minimal $\ell^1$-norm - an effect known as implicit regularization. In this paper, we investigate implicit regularization in diagonal linear neural networks of depth $D\ge 2$ for overparameterized linear regression problems. We focus on analyzing the approximation error between the limit point of gradient flow trajectories and the solution to the $\ell^1$-minimization problem. By deriving tight upper and lower bounds on the approximation error, we precisely characterize how the approximation error depends on the scale of initialization $\alpha$. Our results reveal a qualitative difference between depths: for $D \ge 3$, the error decreases linearly with $\alpha$, whereas for $D=2$, it decreases at rate $\alpha^{1-\varrho}$, where the parameter $\varrho \in [0,1)$ can be explicitly characterized. Interestingly, this parameter is closely linked to so-called null space property constants studied in the sparse recovery literature. We demonstrate the asymptotic tightness of our bounds through explicit examples. Numerical experiments corroborate our theoretical findings and suggest that deeper networks, i.e., $D \ge 3$, may lead to better generalization, particularly for realistic initialization scales.
- [342] arXiv:2506.01147 (cross-list from cs.CL) [pdf, html, other]
-
Title: A Word is Worth 4-bit: Efficient Log Parsing with Binary Coded Decimal RecognitionComments: Pre-print of our accepted paper at IEEE International Conference on Web Services (ICWS 2025). 4 pages, 2 figuresSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
System-generated logs are typically converted into categorical log templates through parsing. These templates are crucial for generating actionable insights in various downstream tasks. However, existing parsers often fail to capture fine-grained template details, leading to suboptimal accuracy and reduced utility in downstream tasks requiring precise pattern identification. We propose a character-level log parser utilizing a novel neural architecture that aggregates character embeddings. Our approach estimates a sequence of binary-coded decimals to achieve highly granular log templates extraction. Our low-resource character-level parser, tested on revised Loghub-2k and a manually annotated industrial dataset, matches LLM-based parsers in accuracy while outperforming semantic parsers in efficiency.
- [343] arXiv:2506.01162 (cross-list from cs.DS) [pdf, html, other]
-
Title: Nearly-Linear Time Private Hypothesis Selection with the Optimal Approximation FactorComments: 33 pagesSubjects: Data Structures and Algorithms (cs.DS); Cryptography and Security (cs.CR); Machine Learning (cs.LG); Machine Learning (stat.ML)
Estimating the density of a distribution from its samples is a fundamental problem in statistics. Hypothesis selection addresses the setting where, in addition to a sample set, we are given $n$ candidate distributions -- referred to as hypotheses -- and the goal is to determine which one best describes the underlying data distribution. This problem is known to be solvable very efficiently, requiring roughly $O(\log n)$ samples and running in $\tilde{O}(n)$ time. The quality of the output is measured via the total variation distance to the unknown distribution, and the approximation factor of the algorithm determines how large this distance is compared to the optimal distance achieved by the best candidate hypothesis. It is known that $\alpha = 3$ is the optimal approximation factor for this problem. We study hypothesis selection under the constraint of differential privacy. We propose a differentially private algorithm in the central model that runs in nearly-linear time with respect to the number of hypotheses, achieves the optimal approximation factor, and incurs only a modest increase in sample complexity, which remains polylogarithmic in $n$. This resolves an open question posed by [Bun, Kamath, Steinke, Wu, NeurIPS 2019]. Prior to our work, existing upper bounds required quadratic time.
- [344] arXiv:2506.01166 (cross-list from cs.AR) [pdf, html, other]
-
Title: VUSA: Virtually Upscaled Systolic Array Architecture to Exploit Unstructured Sparsity in AI AccelerationComments: Preprint accepted for publication at MOCAST 2025. Submitted for possible publication in IEEE XploreSubjects: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Leveraging high degrees of unstructured sparsity is a promising approach to enhance the efficiency of deep neural network DNN accelerators - particularly important for emerging Edge-AI applications. We introduce VUSA, a systolic-array architecture that virtually grows based on the present sparsity to perform larger matrix multiplications with the same number of physical multiply-accumulate MAC units. The proposed architecture achieves saving by 37% and 68% in area and power efficiency, respectively, at the same peak-performance, compared to a baseline systolic array architecture in a commercial 16-nm technology. Still, the proposed architecture supports acceleration for any DNN with any sparsity - even no sparsity at all. Thus, the proposed architecture is application-independent, making it viable for general-purpose AI acceleration.
- [345] arXiv:2506.01173 (cross-list from cs.DB) [pdf, html, other]
-
Title: SIFBench: An Extensive Benchmark for Fatigue AnalysisSubjects: Databases (cs.DB); Machine Learning (cs.LG)
Fatigue-induced crack growth is a leading cause of structural failure across critical industries such as aerospace, civil engineering, automotive, and energy. Accurate prediction of stress intensity factors (SIFs) -- the key parameters governing crack propagation in linear elastic fracture mechanics -- is essential for assessing fatigue life and ensuring structural integrity. While machine learning (ML) has shown great promise in SIF prediction, its advancement has been severely limited by the lack of rich, transparent, well-organized, and high-quality datasets.
To address this gap, we introduce SIFBench, an open-source, large-scale benchmark database designed to support ML-based SIF prediction. SIFBench contains over 5 million different crack and component geometries derived from high-fidelity finite element simulations across 37 distinct scenarios, and provides a unified Python interface for seamless data access and customization. We report baseline results using a range of popular ML models -- including random forests, support vector machines, feedforward neural networks, and Fourier neural operators -- alongside comprehensive evaluation metrics and template code for model training, validation, and assessment. By offering a standardized and scalable resource, SIFBench substantially lowers the entry barrier and fosters the development and application of ML methods in damage tolerance design and predictive maintenance. - [346] arXiv:2506.01189 (cross-list from cs.CV) [pdf, html, other]
-
Title: SVarM: Linear Support Varifold Machines for Classification and Regression on Geometric DataComments: 22 pages, 12 figuresSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Differential Geometry (math.DG); Functional Analysis (math.FA)
Despite progress in the rapidly developing field of geometric deep learning, performing statistical analysis on geometric data--where each observation is a shape such as a curve, graph, or surface--remains challenging due to the non-Euclidean nature of shape spaces, which are defined as equivalence classes under invariance groups. Building machine learning frameworks that incorporate such invariances, notably to shape parametrization, is often crucial to ensure generalizability of the trained models to new observations. This work proposes SVarM to exploit varifold representations of shapes as measures and their duality with test functions $h:\mathbb{R}^n \times S^{n-1} \to \mathbb{R}$. This method provides a general framework akin to linear support vector machines but operating instead over the infinite-dimensional space of varifolds. We develop classification and regression models on shape datasets by introducing a neural network-based representation of the trainable test function $h$. This approach demonstrates strong performance and robustness across various shape graph and surface datasets, achieving results comparable to state-of-the-art methods while significantly reducing the number of trainable parameters.
- [347] arXiv:2506.01197 (cross-list from cs.CL) [pdf, html, other]
-
Title: Incorporating Hierarchical Semantics in Sparse Autoencoder ArchitecturesComments: Code is available at this https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Sparse dictionary learning (and, in particular, sparse autoencoders) attempts to learn a set of human-understandable concepts that can explain variation on an abstract space. A basic limitation of this approach is that it neither exploits nor represents the semantic relationships between the learned concepts. In this paper, we introduce a modified SAE architecture that explicitly models a semantic hierarchy of concepts. Application of this architecture to the internal representations of large language models shows both that semantic hierarchy can be learned, and that doing so improves both reconstruction and interpretability. Additionally, the architecture leads to significant improvements in computational efficiency.
- [348] arXiv:2506.01215 (cross-list from cs.CL) [pdf, html, other]
-
Title: Compress, Gather, and Recompute: REFORMing Long-Context Processing in TransformersWoomin Song, Sai Muralidhar Jayanthi, Srikanth Ronanki, Kanthashree Mysore Sathyendra, Jinwoo Shin, Aram Galstyan, Shubham Katiyar, Sravan Babu BodapatiSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
As large language models increasingly gain popularity in real-world applications, processing extremely long contexts, often exceeding the model's pre-trained context limits, has emerged as a critical challenge. While existing approaches to efficient long-context processing show promise, recurrent compression-based methods struggle with information preservation, whereas random access approaches require substantial memory resources. We introduce REFORM, a novel inference framework that efficiently handles long contexts through a two-phase approach. First, it incrementally processes input chunks while maintaining a compressed KV cache, constructs cross-layer context embeddings, and utilizes early exit strategy for improved efficiency. Second, it identifies and gathers essential tokens via similarity matching and selectively recomputes the KV cache. Compared to baselines, REFORM achieves over 50% and 27% performance gains on RULER and BABILong respectively at 1M context length. It also outperforms baselines on Infinite-Bench and MM-NIAH, demonstrating flexibility across diverse tasks and domains. Additionally, REFORM reduces inference time by 30% and peak memory usage by 5%, achieving both efficiency and superior performance.
- [349] arXiv:2506.01221 (cross-list from eess.IV) [pdf, html, other]
-
Title: Flexible Mixed Precision Quantization for Learned Image CompressionSubjects: Image and Video Processing (eess.IV); Machine Learning (cs.LG)
Despite its improvements in coding performance compared to traditional codecs, Learned Image Compression (LIC) suffers from large computational costs for storage and deployment. Model quantization offers an effective solution to reduce the computational complexity of LIC models. However, most existing works perform fixed-precision quantization which suffers from sub-optimal utilization of resources due to the varying sensitivity to quantization of different layers of a neural network. In this paper, we propose a Flexible Mixed Precision Quantization (FMPQ) method that assigns different bit-widths to different layers of the quantized network using the fractional change in rate-distortion loss as the bit-assignment criterion. We also introduce an adaptive search algorithm which reduces the time-complexity of searching for the desired distribution of quantization bit-widths given a fixed model size. Evaluation of our method shows improved BD-Rate performance under similar model size constraints compared to other works on quantization of LIC models. We have made the source code available at this http URL.
- [350] arXiv:2506.01226 (cross-list from eess.SY) [pdf, html, other]
-
Title: React to Surprises: Stable-by-Design Neural Feedback Control and the Youla-RENSubjects: Systems and Control (eess.SY); Machine Learning (cs.LG)
We study parameterizations of stabilizing nonlinear policies for learning-based control. We propose a structure based on a nonlinear version of the Youla-Kučera parameterization combined with robust neural networks such as the recurrent equilibrium network (REN). The resulting parameterizations are unconstrained, and hence can be searched over with first-order optimization methods, while always ensuring closed-loop stability by construction. We study the combination of (a) nonlinear dynamics, (b) partial observation, and (c) incremental closed-loop stability requirements (contraction and Lipschitzness). We find that with any two of these three difficulties, a contracting and Lipschitz Youla parameter always leads to contracting and Lipschitz closed loops. However, if all three hold, then incremental stability can be lost with exogenous disturbances. Instead, a weaker condition is maintained, which we call d-tube contraction and Lipschitzness. We further obtain converse results showing that the proposed parameterization covers all contracting and Lipschitz closed loops for certain classes of nonlinear systems. Numerical experiments illustrate the utility of our parameterization when learning controllers with built-in stability certificates for: i) ``economic'' rewards without stabilizing effects; ii) short training horizons; and iii) uncertain systems.
- [351] arXiv:2506.01247 (cross-list from cs.CV) [pdf, html, other]
-
Title: Visual Sparse Steering: Improving Zero-shot Image Classification with Sparsity Guided Steering VectorsSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Steering vision foundation models at inference time without retraining or access to large labeled datasets is a desirable yet challenging objective, particularly in dynamic or resource-constrained settings. In this paper, we introduce Visual Sparse Steering (VS2), a lightweight, test-time method that guides vision models using steering vectors derived from sparse features learned by top-$k$ Sparse Autoencoders without requiring contrastive data. Specifically, VS2 surpasses zero-shot CLIP by 4.12% on CIFAR-100, 1.08% on CUB-200, and 1.84% on Tiny-ImageNet. We further propose VS2++, a retrieval-augmented variant that selectively amplifies relevant sparse features using pseudo-labeled neighbors at inference time. With oracle positive/negative sets, VS2++ achieves absolute top-1 gains over CLIP zero-shot of up to 21.44% on CIFAR-100, 7.08% on CUB-200, and 20.47% on Tiny-ImageNet. Interestingly, VS2 and VS2++ raise per-class accuracy by up to 25% and 38%, respectively, showing that sparse steering benefits specific classes by disambiguating visually or taxonomically proximate categories rather than providing a uniform boost. Finally, to better align the sparse features learned through the SAE reconstruction task with those relevant for downstream performance, we propose Prototype-Aligned Sparse Steering (PASS). By incorporating a prototype-alignment loss during SAE training, using labels only during training while remaining fully test-time unsupervised, PASS consistently, though modestly, outperforms VS2, achieving a 6.12% gain over VS2 only on CIFAR-100 with ViT-B/32.
- [352] arXiv:2506.01256 (cross-list from eess.AS) [pdf, html, other]
-
Title: Confidence intervals for forced alignment boundaries using model ensemblesComments: submitted for publication; 7 pages, 1 figureSubjects: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Machine Learning (cs.LG); Sound (cs.SD)
Forced alignment is a common tool to align audio with orthographic and phonetic transcriptions. Most forced alignment tools provide only a single estimate of a boundary. The present project introduces a method of deriving confidence intervals for these boundaries using a neural network ensemble technique. Ten different segment classifier neural networks were previously trained, and the alignment process is repeated with each model. The alignment ensemble is then used to place the boundary at the median of the boundaries in the ensemble, and 97.85% confidence intervals are constructed using order statistics. On the Buckeye and TIMIT corpora, the ensemble boundaries show a slight improvement over using just a single model. The confidence intervals are incorporated into Praat TextGrids using a point tier, and they are also output as a table for researchers to analyze separately as diagnostics or to incorporate uncertainty into their analyses.
- [353] arXiv:2506.01267 (cross-list from stat.ML) [pdf, html, other]
-
Title: Adversarial learning for nonparametric regression: Minimax rate and adaptive estimationSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST); Methodology (stat.ME)
Despite tremendous advancements of machine learning models and algorithms in various application domains, they are known to be vulnerable to subtle, natural or intentionally crafted perturbations in future input data, known as adversarial attacks. While numerous adversarial learning methods have been proposed, fundamental questions about their statistical optimality in robust loss remain largely unanswered. In particular, the minimax rate of convergence and the construction of rate-optimal estimators under future $X$-attacks are yet to be worked out.
In this paper, we address this issue in the context of nonparametric regression, under suitable assumptions on the smoothness of the regression function and the geometric structure of the input perturbation set. We first establish the minimax rate of convergence under adversarial $L_q$-risks with $1 \leq q \leq \infty$ and propose a piecewise local polynomial estimator that achieves the minimax optimality. The established minimax rate elucidates how the smoothness level and perturbation magnitude affect the fundamental limit of adversarial learning under future $X$-attacks. Furthermore, we construct a data-driven adaptive estimator that is shown to achieve, within a logarithmic factor, the optimal rate across a broad scale of nonparametric and adversarial classes. - [354] arXiv:2506.01268 (cross-list from cs.AI) [pdf, html, other]
-
Title: CleanS2S: Single-file Framework for Proactive Speech-to-Speech InteractionSubjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
CleanS2S is a framework for human-like speech-to-speech interaction that advances conversational AI through single-file implementation and proactive dialogue capabilities. Our system integrates automatic speech recognition, large language models, and text-to-speech synthesis into a unified pipeline with real-time interruption handling, achieving low transition latency through full-duplex websocket connections and non-blocking I/O. Beyond conventional chatbot paradigms, we pioneer a proactive interaction mechanism, which combines memory systems with Subjective Action Judgement module, enabling five human-like response strategies: interruption, refusal, deflection, silence, and standard response. The memory module dynamically aggregates historical, and contextual data to inform interaction decisions. This approach breaks the rigid turn-based convention by allowing system-initiated dialog control and context-aware response selection. And we propose Action Judgement SFT that assesses input streams for responses strategies. The framework's single-file implementation with atomic configurations offers researchers unprecedented transparency and extensibility for interaction agents. The code of CleanS2S is released at \this https URL.
- [355] arXiv:2506.01299 (cross-list from cs.AI) [pdf, html, other]
-
Title: Scalable In-Context Q-LearningSubjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Recent advancements in language models have demonstrated remarkable in-context learning abilities, prompting the exploration of in-context reinforcement learning (ICRL) to extend the promise to decision domains. Due to involving more complex dynamics and temporal correlations, existing ICRL approaches may face challenges in learning from suboptimal trajectories and achieving precise in-context inference. In the paper, we propose \textbf{S}calable \textbf{I}n-\textbf{C}ontext \textbf{Q}-\textbf{L}earning (\textbf{SICQL}), an innovative framework that harnesses dynamic programming and world modeling to steer ICRL toward efficient reward maximization and task generalization, while retaining the scalability and stability of supervised pretraining. We design a prompt-based multi-head transformer architecture that simultaneously predicts optimal policies and in-context value functions using separate heads. We pretrain a generalized world model to capture task-relevant information, enabling the construction of a compact prompt that facilitates fast and precise in-context inference. During training, we perform iterative policy improvement by fitting a state value function to an upper-expectile of the Q-function, and distill the in-context value functions into policy extraction using advantage-weighted regression. Extensive experiments across a range of discrete and continuous environments show consistent performance gains over various types of baselines, especially when learning from suboptimal data. Our code is available at this https URL
- [356] arXiv:2506.01308 (cross-list from cs.CL) [pdf, other]
-
Title: A Platform for Investigating Public Health Content with Efficient Concern ClassificationComments: 19 pages, 15 figuresSubjects: Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
A recent rise in online content expressing concerns with public health initiatives has contributed to already stalled uptake of preemptive measures globally. Future public health efforts must attempt to understand such content, what concerns it may raise among readers, and how to effectively respond to it. To this end, we present ConcernScope, a platform that uses a teacher-student framework for knowledge transfer between large language models and light-weight classifiers to quickly and effectively identify the health concerns raised in a text corpus. The platform allows uploading massive files directly, automatically scraping specific URLs, and direct text editing. ConcernScope is built on top of a taxonomy of public health concerns. Intended for public health officials, we demonstrate several applications of this platform: guided data exploration to find useful examples of common concerns found in online community datasets, identification of trends in concerns through an example time series analysis of 186,000 samples, and finding trends in topic frequency before and after significant events.
- [357] arXiv:2506.01324 (cross-list from stat.ML) [pdf, html, other]
-
Title: Near-Optimal Clustering in Mixture of Markov ChainsComments: 36 pagesSubjects: Machine Learning (stat.ML); Information Theory (cs.IT); Machine Learning (cs.LG); Probability (math.PR)
We study the problem of clustering $T$ trajectories of length $H$, each generated by one of $K$ unknown ergodic Markov chains over a finite state space of size $S$. The goal is to accurately group trajectories according to their underlying generative model. We begin by deriving an instance-dependent, high-probability lower bound on the clustering error rate, governed by the weighted KL divergence between the transition kernels of the chains. We then present a novel two-stage clustering algorithm. In Stage~I, we apply spectral clustering using a new injective Euclidean embedding for ergodic Markov chains -- a contribution of independent interest that enables sharp concentration results. Stage~II refines the initial clusters via a single step of likelihood-based reassignment. Our method achieves a near-optimal clustering error with high probability, under the conditions $H = \tilde{\Omega}(\gamma_{\mathrm{ps}}^{-1} (S^2 \vee \pi_{\min}^{-1}))$ and $TH = \tilde{\Omega}(\gamma_{\mathrm{ps}}^{-1} S^2 )$, where $\pi_{\min}$ is the minimum stationary probability of a state across the $K$ chains and $\gamma_{\mathrm{ps}}$ is the minimum pseudo-spectral gap. These requirements provide significant improvements, if not at least comparable, to the state-of-the-art guarantee (Kausik et al., 2023), and moreover, our algorithm offers a key practical advantage: unlike existing approach, it requires no prior knowledge of model-specific quantities (e.g., separation between kernels or visitation probabilities). We conclude by discussing the inherent gap between our upper and lower bounds, providing insights into the unique structure of this clustering problem.
- [358] arXiv:2506.01347 (cross-list from cs.CL) [pdf, html, other]
-
Title: The Surprising Effectiveness of Negative Reinforcement in LLM ReasoningSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Reinforcement learning with verifiable rewards (RLVR) is a promising approach for training language models (LMs) on reasoning tasks that elicit emergent long chains of thought (CoTs). Unlike supervised learning, it updates the model using both correct and incorrect samples via policy gradients. To better understand its mechanism, we decompose the learning signal into reinforcing correct responses and penalizing incorrect ones, referred to as Positive and Negative Sample Reinforcement (PSR and NSR), respectively. We train Qwen2.5-Math-7B and Qwen3-4B on a mathematical reasoning dataset and uncover a surprising result: training with only negative samples -- without reinforcing correct responses -- can be highly effective: it consistently improves performance over the base model across the entire Pass@$k$ spectrum ($k$ up to $256$), often matching or surpassing PPO and GRPO. In contrast, reinforcing only correct responses improves Pass@$1$ but degrades performance at higher $k$, due to reduced diversity. These inference-scaling trends highlight that solely penalizing incorrect responses may contribute more to performance than previously recognized. Through gradient analysis, we show that NSR works by suppressing incorrect generations and redistributing probability mass toward other plausible candidates, guided by the model's prior beliefs. It refines the model's existing knowledge rather than introducing entirely new behaviors. Building on this insight, we propose a simple variant of the RL objective that upweights NSR, and show that it consistently improves overall Pass@$k$ performance on MATH, AIME 2025, and AMC23. Our code is available at this https URL.
- [359] arXiv:2506.01353 (cross-list from cs.AI) [pdf, html, other]
-
Title: EgoBrain: Synergizing Minds and Eyes For Human Action UnderstandingNie Lin, Yansen Wang, Dongqi Han, Weibang Jiang, Jingyuan Li, Ryosuke Furuta, Yoichi Sato, Dongsheng LiComments: 21 pages, 12 figuresSubjects: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
The integration of brain-computer interfaces (BCIs), in particular electroencephalography (EEG), with artificial intelligence (AI) has shown tremendous promise in decoding human cognition and behavior from neural signals. In particular, the rise of multimodal AI models have brought new possibilities that have never been imagined before. Here, we present EgoBrain --the world's first large-scale, temporally aligned multimodal dataset that synchronizes egocentric vision and EEG of human brain over extended periods of time, establishing a new paradigm for human-centered behavior analysis. This dataset comprises 61 hours of synchronized 32-channel EEG recordings and first-person video from 40 participants engaged in 29 categories of daily activities. We then developed a muiltimodal learning framework to fuse EEG and vision for action understanding, validated across both cross-subject and cross-environment challenges, achieving an action recognition accuracy of 66.70%. EgoBrain paves the way for a unified framework for brain-computer interface with multiple modalities. All data, tools, and acquisition protocols are openly shared to foster open science in cognitive computing.
- [360] arXiv:2506.01372 (cross-list from cs.AI) [pdf, html, other]
-
Title: AI Scientists Fail Without Strong Implementation CapabilityComments: PositionSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
The emergence of Artificial Intelligence (AI) Scientist represents a paradigm shift in scientific discovery, with large language models (LLMs) taking the lead as the primary executor in the entire scientific workflow from idea generation to experiment implementation. Recent AI Scientist studies demonstrate sufficient capabilities for independent scientific discovery, with the generated research reports gaining acceptance at the ICLR 2025 workshop and ACL 2025, arguing that a human-level AI Scientist, capable of uncovering phenomena previously unknown to humans, may be imminent. Despite this substantial progress, AI Scientist has yet to produce a groundbreaking achievement in the domain of computer science on par with automated scientific tools. Based on extensive quantitative evidence from existing benchmarks in complex engineering tasks and a systematic evaluation assess 28 research papers generated by five advanced AI Scientist systems, we argue that \textbf{the fundamental bottleneck for AI Scientists lies in their capability to execute the requisite verification procedures.} Current AI Scientist systems lack the execution capabilities needed to execute rigorous experiments and produce high-quality scientific papers. To better illustrate the root cause of this \textbf{implementation gap}, we provide an in-depth discussion on the fundamental limitations of AI Scientist. This position paper aims to call for the participants in the community to bridge the implementation gap.
- [361] arXiv:2506.01378 (cross-list from eess.SP) [pdf, html, other]
-
Title: From Turbulence to Tranquility: AI-Driven Low-Altitude NetworkKürşat Tekbıyık, Amir Hossein Fahim Raouf, İsmail Güvenç, Mingzhe Chen, Güneş Karabulut Kurt, Antoine Lesage-LandrySubjects: Signal Processing (eess.SP); Machine Learning (cs.LG)
Low Altitude Economy (LAE) networks own transformative potential in urban mobility, emergency response, and aerial logistics. However, these networks face significant challenges in spectrum management, interference mitigation, and real-time coordination across dynamic and resource-constrained environments. After addressing these challenges, this study explores three core elements for enabling intelligent LAE networks as follows machine learning-based spectrum sensing and coexistence, artificial intelligence (AI)-optimized resource allocation and trajectory planning, and testbed-driven validation and standardization. We highlight how federated and reinforcement learning techniques support decentralized, adaptive decision-making under mobility and energy constraints. In addition, we discuss the role of real-world platforms such as AERPAW in bridging the gap between simulation and deployment and enabling iterative system refinement under realistic conditions. This study aims to provide a forward-looking roadmap toward developing efficient and interoperable AI-driven LAE ecosystems.
- [362] arXiv:2506.01412 (cross-list from cs.CR) [pdf, html, other]
-
Title: System Calls for Malware Detection and Classification: Methodologies and ApplicationsSubjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
As malware continues to become more complex and harder to detect, Malware Analysis needs to continue to evolve to stay one step ahead. One promising key area approach focuses on using system calls and API Calls, the core communication between user applications and the operating system and their kernels. These calls provide valuable insight into how software or programs behaves, making them an useful tool for spotting suspicious or harmful activity of programs and software. This chapter takes a deep down look at how system calls are used in malware detection and classification, covering techniques like static and dynamic analysis, as well as sandboxing. By combining these methods with advanced techniques like machine learning, statistical analysis, and anomaly detection, researchers can analyze system call patterns to tell the difference between normal and malicious behavior. The chapter also explores how these techniques are applied across different systems, including Windows, Linux, and Android, while also looking at the ways sophisticated malware tries to evade detection.
- [363] arXiv:2506.01413 (cross-list from cs.CV) [pdf, html, other]
-
Title: Incentivizing Reasoning for Advanced Instruction-Following of Large Language ModelsComments: 10 pages of main body, 3 tables, 5 figures, 40 pages of appendixSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Existing large language models (LLMs) face challenges of following complex instructions, especially when multiple constraints are present and organized in paralleling, chaining, and branching structures. One intuitive solution, namely chain-of-thought (CoT), is expected to universally improve capabilities of LLMs. However, we find that the vanilla CoT exerts a negative impact on performance due to its superficial reasoning pattern of simply paraphrasing the instructions. It fails to peel back the compositions of constraints for identifying their relationship across hierarchies of types and dimensions. To this end, we propose a systematic method to boost LLMs in dealing with complex instructions via incentivizing reasoning for test-time compute scaling. First, we stem from the decomposition of complex instructions under existing taxonomies and propose a reproducible data acquisition method. Second, we exploit reinforcement learning (RL) with verifiable rule-centric reward signals to cultivate reasoning specifically for instruction following. We address the shallow, non-essential nature of reasoning under complex instructions via sample-wise contrast for superior CoT enforcement. We also exploit behavior cloning of experts to facilitate steady distribution shift from fast-thinking LLMs to skillful reasoners. Extensive evaluations on seven comprehensive benchmarks confirm the validity of the proposed method, where a 1.5B LLM achieves 11.74% gains with performance comparable to a 8B LLM. Codes and data are available at this https URL.
- [364] arXiv:2506.01420 (cross-list from cs.CL) [pdf, other]
-
Title: Self-Refining Language Model Anonymizers via Adversarial DistillationComments: PreprintSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Large language models (LLMs) are increasingly used in sensitive domains, where their ability to infer personal data from seemingly benign text poses emerging privacy risks. While recent LLM-based anonymization methods help mitigate such risks, they often rely on proprietary models (e.g., GPT-4), raising concerns about cost and the potential exposure of sensitive data to untrusted external systems. To address this, we introduce SElf-refining Anonymization with Language model (SEAL), a novel distillation framework for training small language models (SLMs) to perform effective anonymization without relying on external costly models at inference time. We leverage adversarial interactions between an LLM anonymizer and an inference model to collect trajectories of anonymized texts and inferred attributes, which are used to distill anonymization, adversarial inference, and utility evaluation capabilities into SLMs via supervised fine-tuning and preference learning. The resulting models learn to both anonymize text and critique their outputs, enabling iterative improvement of anonymization quality via self-refinement. Experiments on SynthPAI, a dataset of synthetic personal profiles and text comments, demonstrate that SLMs trained with SEAL achieve substantial improvements in anonymization capabilities. Notably, 8B models attain a privacy-utility trade-off comparable to that of the GPT-4 anonymizer and, with self-refinement, even surpass it in terms of privacy. These results show the effectiveness of our adversarial distillation framework in training SLMs as efficient anonymizers. To facilitate further research, we release the full dataset used in our experiments.
- [365] arXiv:2506.01456 (cross-list from q-bio.GN) [pdf, other]
-
Title: GenDMR: A dynamic multimodal role-swapping network for identifying risk gene phenotypesLina Qin, Cheng Zhu, Chuqi Zhou, Yukun Huang, Jiayi Zhu, Ping Liang, Jinju Wang, Yixing Huang, Cheng Luo, Dezhong Yao, Ying TanComments: 31 pages, 9 figuresSubjects: Genomics (q-bio.GN); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC)
Recent studies have shown that integrating multimodal data fusion techniques for imaging and genetic features is beneficial for the etiological analysis and predictive diagnosis of Alzheimer's disease (AD). However, there are several critical flaws in current deep learning methods. Firstly, there has been insufficient discussion and exploration regarding the selection and encoding of genetic information. Secondly, due to the significantly superior classification value of AD imaging features compared to genetic features, many studies in multimodal fusion emphasize the strengths of imaging features, actively mitigating the influence of weaker features, thereby diminishing the learning of the unique value of genetic features. To address this issue, this study proposes the dynamic multimodal role-swapping network (GenDMR). In GenDMR, we develop a novel approach to encode the spatial organization of single nucleotide polymorphisms (SNPs), enhancing the representation of their genomic context. Additionally, to adaptively quantify the disease risk of SNPs and brain region, we propose a multi-instance attention module to enhance model interpretability. Furthermore, we introduce a dominant modality selection module and a contrastive self-distillation module, combining them to achieve a dynamic teacher-student role exchange mechanism based on dominant and auxiliary modalities for bidirectional co-updating of different modal data. Finally, GenDMR achieves state-of-the-art performance on the ADNI public dataset and visualizes attention to different SNPs, focusing on confirming 12 potential high-risk genes related to AD, including the most classic APOE and recently highlighted significant risk genes. This demonstrates GenDMR's interpretable analytical capability in exploring AD genetic features, providing new insights and perspectives for the development of multimodal data fusion techniques.
- [366] arXiv:2506.01493 (cross-list from cs.CV) [pdf, html, other]
-
Title: Efficiency without Compromise: CLIP-aided Text-to-Image GANs with Increased DiversityComments: Accepted at IJCNN 2025Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Recently, Generative Adversarial Networks (GANs) have been successfully scaled to billion-scale large text-to-image datasets. However, training such models entails a high training cost, limiting some applications and research usage. To reduce the cost, one promising direction is the incorporation of pre-trained models. The existing method of utilizing pre-trained models for a generator significantly reduced the training cost compared with the other large-scale GANs, but we found the model loses the diversity of generation for a given prompt by a large margin. To build an efficient and high-fidelity text-to-image GAN without compromise, we propose to use two specialized discriminators with Slicing Adversarial Networks (SANs) adapted for text-to-image tasks. Our proposed model, called SCAD, shows a notable enhancement in diversity for a given prompt with better sample fidelity. We also propose to use a metric called Per-Prompt Diversity (PPD) to evaluate the diversity of text-to-image models quantitatively. SCAD achieved a zero-shot FID competitive with the latest large-scale GANs at two orders of magnitude less training cost.
- [367] arXiv:2506.01497 (cross-list from cs.NE) [pdf, html, other]
-
Title: SpiceMixer - Netlist-Level Circuit EvolutionStefan Uhlich, Andrea Bonetti, Arun Venkitaraman, Chia-Yu Hsieh, Mustafa Emre Gürsoy, Ryoga Matsuo, Lorenzo ServadeiSubjects: Neural and Evolutionary Computing (cs.NE); Hardware Architecture (cs.AR); Machine Learning (cs.LG)
This paper introduces SpiceMixer, a genetic algorithm developed to synthesize novel analog circuits by evolving SPICE netlists. Unlike conventional methods, SpiceMixer operates directly on netlist lines, enabling compatibility with any component or subcircuit type and supporting general-purpose genetic operations. By using a normalized netlist format, the algorithm enhances the effectiveness of its genetic operators: crossover, mutation, and pruning. We show that SpiceMixer achieves superior performance in synthesizing standard cells (inverter, two-input NAND, and latch) and in designing an analog classifier circuit for the Iris dataset, reaching an accuracy of 89% on the test set. Across all evaluated tasks, SpiceMixer consistently outperforms existing synthesis methods.
- [368] arXiv:2506.01566 (cross-list from cs.PF) [pdf, html, other]
-
Title: FlexiSAGA: A Flexible Systolic Array GEMM Accelerator for Sparse and Dense ProcessingMika Markus Müller, Konstantin Lübeck, Alexander Louis-Ferdinand Jung, Jannik Steinmetz, Oliver BringmannComments: Accepted Version for: SAMOS XXVSubjects: Performance (cs.PF); Artificial Intelligence (cs.AI); Hardware Architecture (cs.AR); Machine Learning (cs.LG)
Artificial Intelligence (AI) algorithms, such as Deep Neural Networks (DNNs), have become an important tool for a wide range of applications, from computer vision to natural language processing. However, the computational complexity of DNN inference poses a significant challenge, particularly for processing on resource-constrained edge devices. One promising approach to address this challenge is the exploitation of sparsity in DNN operator weights.
In this work, we present FlexiSAGA, an architecturally configurable and dataflow-flexible AI hardware accelerator for the sparse and dense processing of general matrix multiplications (GEMMs). FlexiSAGA supports seven different sparse and dense dataflows, enabling efficient processing of resource intensive DNN operators. Additionally, we propose a DNN pruning method specifically tailored towards the FlexiSAGA architecture, allowing for near-optimal processing of dense and sparse convolution and fully-connected operators, facilitating a DNN/HW co-design flow. Our results show a whole DNN sparse-over-dense inference speedup ranging from 1.41 up to 4.28, outperforming commercial and literature-reported accelerator platforms. - [369] arXiv:2506.01586 (cross-list from cs.CV) [pdf, html, other]
-
Title: Multi-Modal Dataset Distillation in the WildSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Recent multi-modal models have shown remarkable versatility in real-world applications. However, their rapid development encounters two critical data challenges. First, the training process requires large-scale datasets, leading to substantial storage and computational costs. Second, these data are typically web-crawled with inevitable noise, i.e., partially mismatched pairs, severely degrading model performance. To these ends, we propose Multi-modal dataset Distillation in the Wild, i.e., MDW, the first framework to distill noisy multi-modal datasets into compact clean ones for effective and efficient model training. Specifically, MDW introduces learnable fine-grained correspondences during distillation and adaptively optimizes distilled data to emphasize correspondence-discriminative regions, thereby enhancing distilled data's information density and efficacy. Moreover, to capture robust cross-modal correspondence prior knowledge from real data, MDW proposes dual-track collaborative learning to avoid the risky data noise, alleviating information loss with certifiable noise tolerance. Extensive experiments validate MDW's theoretical and empirical efficacy with remarkable scalability, surpassing prior methods by over 15% across various compression ratios, highlighting its appealing practicality for applications with diverse efficacy and resource needs.
- [370] arXiv:2506.01602 (cross-list from cs.CL) [pdf, html, other]
-
Title: MMD-Sense-Analysis: Word Sense Detection Leveraging Maximum Mean DiscrepancySubjects: Computation and Language (cs.CL); Machine Learning (cs.LG); Machine Learning (stat.ML)
Word sense analysis is an essential analysis work for interpreting the linguistic and social backgrounds. The word sense change detection is a task of identifying and interpreting shifts in word meanings over time. This paper proposes MMD-Sense-Analysis, a novel approach that leverages Maximum Mean Discrepancy (MMD) to select semantically meaningful variables and quantify changes across time periods. This method enables both the identification of words undergoing sense shifts and the explanation of their evolution over multiple historical periods. To my knowledge, this is the first application of MMD to word sense change detection. Empirical assessment results demonstrate the effectiveness of the proposed approach.
- [371] arXiv:2506.01608 (cross-list from cs.CV) [pdf, html, other]
-
Title: EPFL-Smart-Kitchen-30: Densely annotated cooking dataset with 3D kinematics to challenge video and language modelsAndy Bonnetto, Haozhe Qi, Franklin Leong, Matea Tashkovska, Mahdi Rad, Solaiman Shokur, Friedhelm Hummel, Silvestro Micera, Marc Pollefeys, Alexander MathisComments: Code and data at: this https URLSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Other Quantitative Biology (q-bio.OT)
Understanding behavior requires datasets that capture humans while carrying out complex tasks. The kitchen is an excellent environment for assessing human motor and cognitive function, as many complex actions are naturally exhibited in kitchens from chopping to cleaning. Here, we introduce the EPFL-Smart-Kitchen-30 dataset, collected in a noninvasive motion capture platform inside a kitchen environment. Nine static RGB-D cameras, inertial measurement units (IMUs) and one head-mounted HoloLens~2 headset were used to capture 3D hand, body, and eye movements. The EPFL-Smart-Kitchen-30 dataset is a multi-view action dataset with synchronized exocentric, egocentric, depth, IMUs, eye gaze, body and hand kinematics spanning 29.7 hours of 16 subjects cooking four different recipes. Action sequences were densely annotated with 33.78 action segments per minute. Leveraging this multi-modal dataset, we propose four benchmarks to advance behavior understanding and modeling through 1) a vision-language benchmark, 2) a semantic text-to-motion generation benchmark, 3) a multi-modal action recognition benchmark, 4) a pose-based action segmentation benchmark. We expect the EPFL-Smart-Kitchen-30 dataset to pave the way for better methods as well as insights to understand the nature of ecologically-valid human behavior. Code and data are available at this https URL
- [372] arXiv:2506.01618 (cross-list from eess.AS) [pdf, html, other]
-
Title: Unsupervised Rhythm and Voice Conversion to Improve ASR on Dysarthric SpeechComments: Accepted at Interspeech 2025Subjects: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Sound (cs.SD)
Automatic speech recognition (ASR) systems struggle with dysarthric speech due to high inter-speaker variability and slow speaking rates. To address this, we explore dysarthric-to-healthy speech conversion for improved ASR performance. Our approach extends the Rhythm and Voice (RnV) conversion framework by introducing a syllable-based rhythm modeling method suited for dysarthric speech. We assess its impact on ASR by training LF-MMI models and fine-tuning Whisper on converted speech. Experiments on the Torgo corpus reveal that LF-MMI achieves significant word error rate reductions, especially for more severe cases of dysarthria, while fine-tuning Whisper on converted data has minimal effect on its performance. These results highlight the potential of unsupervised rhythm and voice conversion for dysarthric ASR. Code available at: this https URL
- [373] arXiv:2506.01622 (cross-list from cs.AI) [pdf, other]
-
Title: General agents need world modelsComments: Accepted ICML 2025Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Robotics (cs.RO); Machine Learning (stat.ML)
Are world models a necessary ingredient for flexible, goal-directed behaviour, or is model-free learning sufficient? We provide a formal answer to this question, showing that any agent capable of generalizing to multi-step goal-directed tasks must have learned a predictive model of its environment. We show that this model can be extracted from the agent's policy, and that increasing the agents performance or the complexity of the goals it can achieve requires learning increasingly accurate world models. This has a number of consequences: from developing safe and general agents, to bounding agent capabilities in complex environments, and providing new algorithms for eliciting world models from agents.
- [374] arXiv:2506.01623 (cross-list from cs.AI) [pdf, html, other]
-
Title: MAGIK: Mapping to Analogous Goals via Imagination-enabled Knowledge TransferSubjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Humans excel at analogical reasoning - applying knowledge from one task to a related one with minimal relearning. In contrast, reinforcement learning (RL) agents typically require extensive retraining even when new tasks share structural similarities with previously learned ones. In this work, we propose MAGIK, a novel framework that enables RL agents to transfer knowledge to analogous tasks without interacting with the target environment. Our approach leverages an imagination mechanism to map entities in the target task to their analogues in the source domain, allowing the agent to reuse its original policy. Experiments on custom MiniGrid and MuJoCo tasks show that MAGIK achieves effective zero-shot transfer using only a small number of human-labelled examples. We compare our approach to related baselines and highlight how it offers a novel and effective mechanism for knowledge transfer via imagination-based analogy mapping.
- [375] arXiv:2506.01624 (cross-list from cs.AI) [pdf, html, other]
-
Title: Social Cooperation in Conversational AI AgentsComments: 4 pages, RLDM 2025 abstract (Spotlight presentation)Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
The development of AI agents based on large, open-domain language models (LLMs) has paved the way for the development of general-purpose AI assistants that can support human in tasks such as writing, coding, graphic design, and scientific research. A major challenge with such agents is that, by necessity, they are trained by observing relatively short-term interactions with humans. Such models can fail to generalize to long-term interactions, for example, interactions where a user has repeatedly corrected mistakes on the part of the agent. In this work, we argue that these challenges can be overcome by explicitly modeling humans' social intelligence, that is, their ability to build and maintain long-term relationships with other agents whose behavior cannot always be predicted. By mathematically modeling the strategies humans use to communicate and reason about one another over long periods of time, we may be able to derive new game theoretic objectives against which LLMs and future AI agents may be optimized.
- [376] arXiv:2506.01635 (cross-list from cs.RO) [pdf, html, other]
-
Title: Riemannian Time Warping: Multiple Sequence Alignment in Curved SpacesSubjects: Robotics (cs.RO); Machine Learning (cs.LG)
Temporal alignment of multiple signals through time warping is crucial in many fields, such as classification within speech recognition or robot motion learning. Almost all related works are limited to data in Euclidean space. Although an attempt was made in 2011 to adapt this concept to unit quaternions, a general extension to Riemannian manifolds remains absent. Given its importance for numerous applications in robotics and beyond, we introduce Riemannian Time Warping~(RTW). This novel approach efficiently aligns multiple signals by considering the geometric structure of the Riemannian manifold in which the data is embedded. Extensive experiments on synthetic and real-world data, including tests with an LBR iiwa robot, demonstrate that RTW consistently outperforms state-of-the-art baselines in both averaging and classification tasks.
- [377] arXiv:2506.01641 (cross-list from eess.SY) [pdf, html, other]
-
Title: Interpretable reinforcement learning for heat pump control through asymmetric differentiable decision treesComments: 7 pages, 3 figures, conferenceSubjects: Systems and Control (eess.SY); Machine Learning (cs.LG)
In recent years, deep reinforcement learning (DRL) algorithms have gained traction in home energy management systems. However, their adoption by energy management companies remains limited due to the black-box nature of DRL, which fails to provide transparent decision-making feedback. To address this, explainable reinforcement learning (XRL) techniques have emerged, aiming to make DRL decisions more transparent. Among these, soft differential decision tree (DDT) distillation provides a promising approach due to the clear decision rules they are based on, which can be efficiently computed. However, achieving high performance often requires deep, and completely full, trees, which reduces interpretability. To overcome this, we propose a novel asymmetric soft DDT construction method. Unlike traditional soft DDTs, our approach adaptively constructs trees by expanding nodes only when necessary. This improves the efficient use of decision nodes, which require a predetermined depth to construct full symmetric trees, enhancing both interpretability and performance. We demonstrate the potential of asymmetric DDTs to provide transparent, efficient, and high-performing decision-making in home energy management systems.
- [378] arXiv:2506.01646 (cross-list from cs.CL) [pdf, html, other]
-
Title: ESGenius: Benchmarking LLMs on Environmental, Social, and Governance (ESG) and Sustainability KnowledgeChaoyue He, Xin Zhou, Yi Wu, Xinjia Yu, Yan Zhang, Lei Zhang, Di Wang, Shengfei Lyu, Hong Xu, Xiaoqiao Wang, Wei Liu, Chunyan MiaoComments: 37 pages, 8 figures, 11 tablesSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
We introduce ESGenius, a comprehensive benchmark for evaluating and enhancing the proficiency of Large Language Models (LLMs) in Environmental, Social and Governance (ESG) and sustainability-focused question answering. ESGenius comprises two key components: (i) ESGenius-QA, a collection of 1 136 multiple-choice questions generated by LLMs and rigorously validated by domain experts, covering a broad range of ESG pillars and sustainability topics. Each question is systematically linked to its corresponding source text, enabling transparent evaluation and supporting retrieval-augmented generation (RAG) methods; and (ii) ESGenius-Corpus, a meticulously curated repository of 231 foundational frameworks, standards, reports and recommendation documents from seven authoritative sources. Moreover, to fully assess the capabilities and adaptation potential of the model, we implement a rigorous two-stage evaluation protocol -- Zero-Shot and RAG. Extensive experiments across 50 LLMs (ranging from 0.5 B to 671 B parameters) demonstrate that state-of-the-art models achieve only moderate performance in zero-shot settings, with accuracies typically around 55--70\%, highlighting ESGenius's challenging nature for LLMs in interdisciplinary contexts. However, models employing RAG show significant performance improvements, particularly for smaller models. For example, "DeepSeek-R1-Distill-Qwen-14B" improves from 63.82\% (zero-shot) to 80.46\% with RAG. These results underscore the necessity of grounding responses in authoritative sources for enhanced ESG understanding. To the best of our knowledge, ESGenius is the first benchmark curated for LLMs and the relevant enhancement technologies that focuses on ESG and sustainability topics.
- [379] arXiv:2506.01659 (cross-list from cs.NE) [pdf, html, other]
-
Title: Engram Memory Encoding and Retrieval: A Neurocomputational PerspectiveComments: 18 pages, 7 figures, 3 tablesSubjects: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC)
Despite substantial research into the biological basis of memory, the precise mechanisms by which experiences are encoded, stored, and retrieved in the brain remain incompletely understood. A growing body of evidence supports the engram theory, which posits that sparse populations of neurons undergo lasting physical and biochemical changes to support long-term memory. Yet, a comprehensive computational framework that integrates biological findings with mechanistic models remains elusive. This work synthesizes insights from cellular neuroscience and computational modeling to address key challenges in engram research: how engram neurons are identified and manipulated; how synaptic plasticity mechanisms contribute to stable memory traces; and how sparsity promotes efficient, interference-resistant representations. Relevant computational approaches -- such as sparse regularization, engram gating, and biologically inspired architectures like Sparse Distributed Memory and spiking neural networks -- are also examined. Together, these findings suggest that memory efficiency, capacity, and stability emerge from the interaction of plasticity and sparsity constraints. By integrating neurobiological and computational perspectives, this paper provides a comprehensive theoretical foundation for engram research and proposes a roadmap for future inquiry into the mechanisms underlying memory, with implications for the diagnosis and treatment of memory-related disorders.
- [380] arXiv:2506.01662 (cross-list from cs.CY) [pdf, html, other]
-
Title: Explainable AI Systems Must Be Contestable: Here's How to Make It HappenCatarina Moreira, Anna Palatkina, Dacia Braca, Dylan M. Walsh, Peter J. Leihn, Fang Chen, Nina C. HubigSubjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
As AI regulations around the world intensify their focus on system safety, contestability has become a mandatory, yet ill-defined, safeguard. In XAI, "contestability" remains an empty promise: no formal definition exists, no algorithm guarantees it, and practitioners lack concrete guidance to satisfy regulatory requirements. Grounded in a systematic literature review, this paper presents the first rigorous formal definition of contestability in explainable AI, directly aligned with stakeholder requirements and regulatory mandates. We introduce a modular framework of by-design and post-hoc mechanisms spanning human-centered interfaces, technical architectures, legal processes, and organizational workflows. To operationalize our framework, we propose the Contestability Assessment Scale, a composite metric built on more than twenty quantitative criteria. Through multiple case studies across diverse application domains, we reveal where state-of-the-art systems fall short and show how our framework drives targeted improvements. By converting contestability from regulatory theory into a practical framework, our work equips practitioners with the tools to embed genuine recourse and accountability into AI systems.
- [381] arXiv:2506.01666 (cross-list from quant-ph) [pdf, html, other]
-
Title: Synthesis of discrete-continuous quantum circuits with multimodal diffusion modelsComments: Main Text: 10 pages and 5 figures; Appendix: 17 pages, 7 figures and 1 table. Code available at: this https URLSubjects: Quantum Physics (quant-ph); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Efficiently compiling quantum operations remains a major bottleneck in scaling quantum computing. Today's state-of-the-art methods achieve low compilation error by combining search algorithms with gradient-based parameter optimization, but they incur long runtimes and require multiple calls to quantum hardware or expensive classical simulations, making their scaling prohibitive. Recently, machine-learning models have emerged as an alternative, though they are currently restricted to discrete gate sets. Here, we introduce a multimodal denoising diffusion model that simultaneously generates a circuit's structure and its continuous parameters for compiling a target unitary. It leverages two independent diffusion processes, one for discrete gate selection and one for parameter prediction. We benchmark the model over different experiments, analyzing the method's accuracy across varying qubit counts, circuit depths, and proportions of parameterized gates. Finally, by exploiting its rapid circuit generation, we create large datasets of circuits for particular operations and use these to extract valuable heuristics that can help us discover new insights into quantum circuit synthesis.
- [382] arXiv:2506.01685 (cross-list from cs.GT) [pdf, html, other]
-
Title: Geometry Meets Incentives: Sample-Efficient Incentivized Exploration with Linear ContextsSubjects: Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG)
In the incentivized exploration model, a principal aims to explore and learn over time by interacting with a sequence of self-interested agents. It has been recently understood that the main challenge in designing incentive-compatible algorithms for this problem is to gather a moderate amount of initial data, after which one can obtain near-optimal regret via posterior sampling. With high-dimensional contexts, however, this \emph{initial exploration} phase requires exponential sample complexity in some cases, which prevents efficient learning unless initial data can be acquired exogenously. We show that these barriers to exploration disappear under mild geometric conditions on the set of available actions, in which case incentive-compatibility does not preclude regret-optimality. Namely, we consider the linear bandit model with actions in the Euclidean unit ball, and give an incentive-compatible exploration algorithm with sample complexity that scales polynomially with the dimension and other parameters.
- [383] arXiv:2506.01692 (cross-list from cs.AI) [pdf, html, other]
-
Title: A Descriptive and Normative Theory of Human Beliefs in RLHFSubjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Human preferences in RLHF are typically modeled as a function of the human's reward function or corresponding optimal state-action values. In this work, we propose that human beliefs about the capabilities of the agent being trained also play a key role in preference generation. We examine two questions related to this hypothesis, one descriptive and one normative, respectively: Do human labelers' beliefs about agent capabilities affect the preferences that they provide? And what is the ideal set of beliefs about an agent -- and resulting preferences -- for humans to have? We propose a new preference model that incorporates human beliefs and provide a normative theory that bounds the error on the final learned policy based on the \textit{mismatch} between the human's beliefs and an idealized set of beliefs. We then confirm via a human study that beliefs about agent capabilities do, in fact, significantly affect preferences and can be influenced through simple interventions. Additionally, we empirically show through synthetic experiments that it is often suboptimal for human preference labelers to assume agent optimality. Collectively, these results theoretically and empirically demonstrate how reducing the mismatch between human beliefs and agent capabilities can lead to more performant RLHF and point toward new best practices for RLHF practitioners.
- [384] arXiv:2506.01718 (cross-list from stat.ML) [pdf, html, other]
-
Title: Signature Maximum Mean Discrepancy Two-Sample Statistical TestsComments: 45 pages, 19 figuresSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Dynamical Systems (math.DS)
Maximum Mean Discrepancy (MMD) is a widely used concept in machine learning research which has gained popularity in recent years as a highly effective tool for comparing (finite-dimensional) distributions. Since it is designed as a kernel-based method, the MMD can be extended to path space valued distributions using the signature kernel. The resulting signature MMD (sig-MMD) can be used to define a metric between distributions on path space. Similarly to the original use case of the MMD as a test statistic within a two-sample testing framework, the sig-MMD can be applied to determine if two sets of paths are drawn from the same stochastic process. This work is dedicated to understanding the possibilities and challenges associated with applying the sig-MMD as a statistical tool in practice. We introduce and explain the sig-MMD, and provide easily accessible and verifiable examples for its practical use. We present examples that can lead to Type 2 errors in the hypothesis test, falsely indicating that samples have been drawn from the same underlying process (which generally occurs in a limited data setting). We then present techniques to mitigate the occurrence of this type of error.
- [385] arXiv:2506.01755 (cross-list from eess.SY) [pdf, html, other]
-
Title: Data-assimilated model-informed reinforcement learningSubjects: Systems and Control (eess.SY); Machine Learning (cs.LG)
The control of spatio-temporally chaos is challenging because of high dimensionality and unpredictability. Model-free reinforcement learning (RL) discovers optimal control policies by interacting with the system, typically requiring observations of the full physical this http URL practice, sensors often provide only partial and noisy measurements (observations) of the system. The objective of this paper is to develop a framework that enables the control of chaotic systems with partial and noisy observability. The proposed method, data-assimilated model-informed reinforcement learning (DA-MIRL), integrates (i) low-order models to approximate high-dimensional dynamics; (ii) sequential data assimilation to correct the model prediction when observations become available; and (iii) an off-policy actor-critic RL algorithm to adaptively learn an optimal control strategy based on the corrected state estimates. We test DA-MIRL on the spatiotemporally chaotic solutions of the Kuramoto-Sivashinsky equation. We estimate the full state of the environment with (i) a physics-based model, here, a coarse-grained model; and (ii) a data-driven model, here, the control-aware echo state network, which is proposed in this paper. We show that DA-MIRL successfully estimates and suppresses the chaotic dynamics of the environment in real time from partial observations and approximate models. This work opens opportunities for the control of partially observable chaotic systems.
- [386] arXiv:2506.01770 (cross-list from cs.CR) [pdf, html, other]
-
Title: ReGA: Representation-Guided Abstraction for Model-based Safeguarding of LLMsSubjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Software Engineering (cs.SE)
Large Language Models (LLMs) have achieved significant success in various tasks, yet concerns about their safety and security have emerged. In particular, they pose risks in generating harmful content and vulnerability to jailbreaking attacks. To analyze and monitor machine learning models, model-based analysis has demonstrated notable potential in stateful deep neural networks, yet suffers from scalability issues when extending to LLMs due to their vast feature spaces. In this paper, we propose ReGA, a model-based analysis framework with representation-guided abstraction, to safeguard LLMs against harmful prompts and generations. By leveraging safety-critical representations, which are low-dimensional directions emerging in hidden states that indicate safety-related concepts, ReGA effectively addresses the scalability issue when constructing the abstract model for safety modeling. Our comprehensive evaluation shows that ReGA performs sufficiently well in distinguishing between safe and harmful inputs, achieving an AUROC of 0.975 at the prompt level and 0.985 at the conversation level. Additionally, ReGA exhibits robustness to real-world attacks and generalization across different safety perspectives, outperforming existing safeguard paradigms in terms of interpretability and scalability. Overall, ReGA serves as an efficient and scalable solution to enhance LLM safety by integrating representation engineering with model-based abstraction, paving the way for new paradigms to utilize software insights for AI safety. Our code is available at this https URL.
- [387] arXiv:2506.01778 (cross-list from cs.CV) [pdf, html, other]
-
Title: unMORE: Unsupervised Multi-Object Segmentation via Center-Boundary ReasoningComments: ICML 2025. Code and data are available at: this https URLSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Robotics (cs.RO)
We study the challenging problem of unsupervised multi-object segmentation on single images. Existing methods, which rely on image reconstruction objectives to learn objectness or leverage pretrained image features to group similar pixels, often succeed only in segmenting simple synthetic objects or discovering a limited number of real-world objects. In this paper, we introduce unMORE, a novel two-stage pipeline designed to identify many complex objects in real-world images. The key to our approach involves explicitly learning three levels of carefully defined object-centric representations in the first stage. Subsequently, our multi-object reasoning module utilizes these learned object priors to discover multiple objects in the second stage. Notably, this reasoning module is entirely network-free and does not require human labels. Extensive experiments demonstrate that unMORE significantly outperforms all existing unsupervised methods across 6 real-world benchmark datasets, including the challenging COCO dataset, achieving state-of-the-art object segmentation results. Remarkably, our method excels in crowded images where all baselines collapse.
- [388] arXiv:2506.01816 (cross-list from math.OC) [pdf, html, other]
-
Title: An adaptive data sampling strategy for stabilizing dynamical systems via controller inferenceComments: 27 pages, 9 figuresSubjects: Optimization and Control (math.OC); Machine Learning (cs.LG); Dynamical Systems (math.DS); Numerical Analysis (math.NA)
Learning stabilizing controllers from data is an important task in engineering applications; however, collecting informative data is challenging because unstable systems often lead to rapidly growing or erratic trajectories. In this work, we propose an adaptive sampling scheme that generates data while simultaneously stabilizing the system to avoid instabilities during the data collection. Under mild assumptions, the approach provably generates data sets that are informative for stabilization and have minimal size. The numerical experiments demonstrate that controller inference with the novel adaptive sampling approach learns controllers with up to one order of magnitude fewer data samples than unguided data generation. The results show that the proposed approach opens the door to stabilizing systems in edge cases and limit states where instabilities often occur and data collection is inherently difficult.
- [389] arXiv:2506.01820 (cross-list from cs.AI) [pdf, html, other]
-
Title: Fodor and Pylyshyn's Legacy - Still No Human-like Systematic Compositionality in Neural NetworksTim Woydt, Moritz Willig, Antonia Wüst, Lukas Helff, Wolfgang Stammer, Constantin A. Rothkopf, Kristian KerstingSubjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Strong meta-learning capabilities for systematic compositionality are emerging as an important skill for navigating the complex and changing tasks of today's world. However, in presenting models for robust adaptation to novel environments, it is important to refrain from making unsupported claims about the performance of meta-learning systems that ultimately do not stand up to scrutiny. While Fodor and Pylyshyn famously posited that neural networks inherently lack this capacity as they are unable to model compositional representations or structure-sensitive operations, and thus are not a viable model of the human mind, Lake and Baroni recently presented meta-learning as a pathway to compositionality. In this position paper, we critically revisit this claim and highlight limitations in the proposed meta-learning framework for compositionality. Our analysis shows that modern neural meta-learning systems can only perform such tasks, if at all, under a very narrow and restricted definition of a meta-learning setup. We therefore claim that `Fodor and Pylyshyn's legacy' persists, and to date, there is no human-like systematic compositionality learned in neural networks.
- [390] arXiv:2506.01845 (cross-list from eess.AS) [pdf, html, other]
-
Title: On-device Streaming Discrete Speech UnitsComments: Accepted to Interspeech 2025, source code at this https URLSubjects: Audio and Speech Processing (eess.AS); Machine Learning (cs.LG); Sound (cs.SD)
Discrete speech units (DSUs) are derived from clustering the features of self-supervised speech models (S3Ms). DSUs offer significant advantages for on-device streaming speech applications due to their rich phonetic information, high transmission efficiency, and seamless integration with large language models. However, conventional DSU-based approaches are impractical as they require full-length speech input and computationally expensive S3Ms. In this work, we reduce both the attention window and the model size while preserving the effectiveness of DSUs. Our results demonstrate that we can reduce floating-point operations (FLOPs) by 50% with only a relative increase of 6.5% in character error rate (CER) on the ML-SUPERB 1h dataset. These findings highlight the potential of DSUs for real-time speech processing in resource-constrained environments.
- [391] arXiv:2506.01850 (cross-list from cs.CV) [pdf, html, other]
-
Title: MoDA: Modulation Adapter for Fine-Grained Visual Grounding in Instructional MLLMsSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multimedia (cs.MM)
Recently, Multimodal Large Language Models (MLLMs) have demonstrated impressive performance on instruction-following tasks by integrating pretrained visual encoders with large language models (LLMs). However, existing approaches often struggle to ground fine-grained visual concepts in complex scenes. In this paper, we propose MoDA (Modulation Adapter), a lightweight yet effective module designed to refine pre-aligned visual features through instruction-guided modulation. Our approach follows the standard LLaVA training protocol, consisting of a two-stage process: (1) aligning image features to the LLMs input space via a frozen vision encoder and adapter layers, and (2) refining those features using the MoDA adapter during the instructional tuning stage. MoDA employs a Transformer-based cross-attention mechanism to generate a modulation mask over the aligned visual tokens, thereby emphasizing semantically relevant embedding dimensions based on the language instruction. The modulated features are then passed to the LLM for autoregressive language generation. Our experimental evaluation shows that MoDA improves visual grounding and generates more contextually appropriate responses, demonstrating its effectiveness as a general-purpose enhancement for image-based MLLMs.
- [392] arXiv:2506.01882 (cross-list from quant-ph) [pdf, html, other]
-
Title: Learning thermodynamic master equations for open quantum systemsComments: 20 pages, 7 figuresSubjects: Quantum Physics (quant-ph); Machine Learning (cs.LG)
The characterization of Hamiltonians and other components of open quantum dynamical systems plays a crucial role in quantum computing and other applications. Scientific machine learning techniques have been applied to this problem in a variety of ways, including by modeling with deep neural networks. However, the majority of mathematical models describing open quantum systems are linear, and the natural nonlinearities in learnable models have not been incorporated using physical principles. We present a data-driven model for open quantum systems that includes learnable, thermodynamically consistent terms. The trained model is interpretable, as it directly estimates the system Hamiltonian and linear components of coupling to the environment. We validate the model on synthetic two and three-level data, as well as experimental two-level data collected from a quantum device at Lawrence Livermore National Laboratory.
- [393] arXiv:2506.01891 (cross-list from quant-ph) [pdf, html, other]
-
Title: Probing Quantum Spin Systems with Kolmogorov-Arnold Neural Network Quantum StatesComments: 16 pages, 13 figuresSubjects: Quantum Physics (quant-ph); Disordered Systems and Neural Networks (cond-mat.dis-nn); Strongly Correlated Electrons (cond-mat.str-el); Machine Learning (cs.LG)
Neural Quantum States (NQS) are a class of variational wave functions parametrized by neural networks (NNs) to study quantum many-body systems. In this work, we propose SineKAN, the NQS ansatz based on Kolmogorov-Arnold Networks (KANs), to represent quantum mechanical wave functions as nested univariate functions. We show that \sk wavefunction with learnable sinusoidal activation functions can capture the ground state energies, fidelities and various correlation functions of the 1D Transverse-Field Ising model, Anisotropic Heisenberg model, and Antiferromagnetic $J_{1}-J_{2}$ model with different chain lengths. In our study of the $J_1-J_2$ model with $L=100$ sites, we find that the SineKAN model outperforms several previously explored neural quantum state ansätze, including Restricted Boltzmann Machines (RBMs), Long Short-Term Memory models (LSTMs), and Feed-Forward Neural Networks (FFCN), when compared to the results obtained from the Density Matrix Renormalization Group (DMRG) algorithm.
- [394] arXiv:2506.01904 (cross-list from stat.ML) [pdf, html, other]
-
Title: Machine-Learned Sampling of Conditioned Path MeasuresSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Computation (stat.CO)
We propose algorithms for sampling from posterior path measures $P(C([0, T], \mathbb{R}^d))$ under a general prior process. This leverages ideas from (1) controlled equilibrium dynamics, which gradually transport between two path measures, and (2) optimization in $\infty$-dimensional probability space endowed with a Wasserstein metric, which can be used to evolve a density curve under the specified likelihood. The resulting algorithms are theoretically grounded and can be integrated seamlessly with neural networks for learning the target trajectory ensembles, without access to data.
- [395] arXiv:2506.01923 (cross-list from cs.CV) [pdf, html, other]
-
Title: TaxaDiffusion: Progressively Trained Diffusion Model for Fine-Grained Species GenerationSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
We propose TaxaDiffusion, a taxonomy-informed training framework for diffusion models to generate fine-grained animal images with high morphological and identity accuracy. Unlike standard approaches that treat each species as an independent category, TaxaDiffusion incorporates domain knowledge that many species exhibit strong visual similarities, with distinctions often residing in subtle variations of shape, pattern, and color. To exploit these relationships, TaxaDiffusion progressively trains conditioned diffusion models across different taxonomic levels -- starting from broad classifications such as Class and Order, refining through Family and Genus, and ultimately distinguishing at the Species level. This hierarchical learning strategy first captures coarse-grained morphological traits shared by species with common ancestors, facilitating knowledge transfer before refining fine-grained differences for species-level distinction. As a result, TaxaDiffusion enables accurate generation even with limited training samples per species. Extensive experiments on three fine-grained animal datasets demonstrate that outperforms existing approaches, achieving superior fidelity in fine-grained animal image generation. Project page: this https URL
- [396] arXiv:2506.01926 (cross-list from cs.AI) [pdf, other]
-
Title: Large language models can learn and generalize steganographic chain-of-thought under process supervisionJoey Skaf, Luis Ibanez-Lissen, Robert McCarthy, Connor Watts, Vasil Georgiv, Hannes Whittingham, Lorena Gonzalez-Manzano, David Lindner, Cameron Tice, Edward James Young, Puria RadmardComments: 10 pages main text, 3 figures main text, 15 pages supplementary material, 1 figure supplementary material, submitted to NeurIPS 2025Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Chain-of-thought (CoT) reasoning not only enhances large language model performance but also provides critical insights into decision-making processes, marking it as a useful tool for monitoring model intent and planning. By proactively preventing models from acting on CoT indicating misaligned or harmful intent, CoT monitoring can be used to reduce risks associated with deploying models. However, developers may be incentivized to train away the appearance of harmful intent from CoT traces, by either customer preferences or regulatory requirements. Recent works have shown that banning mention of a specific example of reward hacking, which may be done either to make CoT presentable to users or as a naive attempt to prevent the behavior, causes obfuscation of the undesired reasoning traces but the persistence of the undesired behavior. Such obfuscation threatens the reliability of CoT monitoring. However, obfuscation of reasoning can be due to its internalization to latent space computation, or its encoding within the CoT. Here, we provide an extension to these results. First, we show that penalizing the use of specific strings within load-bearing reasoning traces causes models to substitute alternative strings. Crucially, this does not alter the underlying method by which the model performs the task, demonstrating that the model can learn to steganographically encode its reasoning. We further demonstrate that models can generalize an encoding scheme. When the penalized strings belong to an overarching class, the model learns not only to substitute strings seen in training, but also develops a general encoding scheme for all members of the class which it can apply to held-out testing strings.
- [397] arXiv:2506.01928 (cross-list from cs.CL) [pdf, html, other]
-
Title: Esoteric Language ModelsSubham Sekhar Sahoo, Zhihan Yang, Yash Akhauri, Johnna Liu, Deepansha Singh, Zhoujun Cheng, Zhengzhong Liu, Eric Xing, John Thickstun, Arash VahdatSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Diffusion-based language models offer a compelling alternative to autoregressive (AR) models by enabling parallel and controllable generation. Among this family of models, Masked Diffusion Models (MDMs) achieve the strongest performance but still underperform AR models in perplexity and lack key inference-time efficiency features--most notably, KV caching. In this work, we introduce Eso-LMs, a new family of models that fuses AR and MDM paradigms, enabling smooth interpolation between their perplexities while overcoming their respective limitations. Eso-LMs set a new state of the art on standard language modeling benchmarks. Crucially, we are the **first to introduce KV caching for MDMs** while preserving parallel generation, significantly improving inference efficiency. Combined with an optimized sampling schedule, our method achieves up to **65x** faster inference than standard MDMs and **4x** faster inference than prior semi-autoregressive approaches. We provide the code and model checkpoints on the project page: [this http URL](this http URL)
- [398] arXiv:2506.01929 (cross-list from cs.GR) [pdf, html, other]
-
Title: Image Generation from Contextually-Contradictory PromptsComments: Project page: this https URLSubjects: Graphics (cs.GR); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Text-to-image diffusion models excel at generating high-quality, diverse images from natural language prompts. However, they often fail to produce semantically accurate results when the prompt contains concept combinations that contradict their learned priors. We define this failure mode as contextual contradiction, where one concept implicitly negates another due to entangled associations learned during training. To address this, we propose a stage-aware prompt decomposition framework that guides the denoising process using a sequence of proxy prompts. Each proxy prompt is constructed to match the semantic content expected to emerge at a specific stage of denoising, while ensuring contextual coherence. To construct these proxy prompts, we leverage a large language model (LLM) to analyze the target prompt, identify contradictions, and generate alternative expressions that preserve the original intent while resolving contextual conflicts. By aligning prompt information with the denoising progression, our method enables fine-grained semantic control and accurate image generation in the presence of contextual contradictions. Experiments across a variety of challenging prompts show substantial improvements in alignment to the textual prompt.
- [399] arXiv:2506.01936 (cross-list from cs.GT) [pdf, html, other]
-
Title: Should Decision-Makers Reveal Classifiers in Online Strategic Classification?Subjects: Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG)
Strategic classification addresses a learning problem where a decision-maker implements a classifier over agents who may manipulate their features in order to receive favorable predictions. In the standard model of online strategic classification, in each round, the decision-maker implements and publicly reveals a classifier, after which agents perfectly best respond based on this knowledge. However, in practice, whether to disclose the classifier is often debated -- some decision-makers believe that hiding the classifier can prevent misclassification errors caused by manipulation.
In this paper, we formally examine how limiting the agents' access to the current classifier affects the decision-maker's performance. Specifically, we consider an extended online strategic classification setting where agents lack direct knowledge about the current classifier and instead manipulate based on a weighted average of historically implemented classifiers. Our main result shows that in this setting, the decision-maker incurs $(1-\gamma)^{-1}$ or $k_{\text{in}}$ times more mistakes compared to the full-knowledge setting, where $k_{\text{in}}$ is the maximum in-degree of the manipulation graph (representing how many distinct feature vectors can be manipulated to appear as a single one), and $\gamma$ is the discount factor indicating agents' memory of past classifiers. Our results demonstrate how withholding access to the classifier can backfire and degrade the decision-maker's performance in online strategic classification. - [400] arXiv:2506.01939 (cross-list from cs.CL) [pdf, other]
-
Title: Beyond the 80/20 Rule: High-Entropy Minority Tokens Drive Effective Reinforcement Learning for LLM ReasoningShenzhi Wang, Le Yu, Chang Gao, Chujie Zheng, Shixuan Liu, Rui Lu, Kai Dang, Xionghui Chen, Jianxin Yang, Zhenru Zhang, Yuqiong Liu, An Yang, Andrew Zhao, Yang Yue, Shiji Song, Bowen Yu, Gao Huang, Junyang LinComments: 25 pages, 17 figures, 2 tablesSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Reinforcement Learning with Verifiable Rewards (RLVR) has emerged as a powerful approach to enhancing the reasoning capabilities of Large Language Models (LLMs), while its mechanisms are not yet well understood. In this work, we undertake a pioneering exploration of RLVR through the novel perspective of token entropy patterns, comprehensively analyzing how different tokens influence reasoning performance. By examining token entropy patterns in Chain-of-Thought (CoT) reasoning, we observe that only a small fraction of tokens exhibit high entropy, and these tokens act as critical forks that steer the model toward diverse reasoning pathways. Furthermore, studying how entropy patterns evolve during RLVR training reveals that RLVR largely adheres to the base model's entropy patterns, primarily adjusting the entropy of high-entropy tokens. These findings highlight the significance of high-entropy tokens (i.e., forking tokens) to RLVR. We ultimately improve RLVR by restricting policy gradient updates to forking tokens and uncover a finding even beyond the 80/20 rule: utilizing only 20% of the tokens while maintaining performance comparable to full-gradient updates on the Qwen3-8B base model and significantly surpassing full-gradient updates on the Qwen3-32B (+11.04 on AIME'25 and +7.71 on AIME'24) and Qwen3-14B (+4.79 on AIME'25 and +5.21 on AIME'24) base models, highlighting a strong scaling trend. In contrast, training exclusively on the 80% lowest-entropy tokens leads to a marked decline in performance. These findings indicate that the efficacy of RLVR primarily arises from optimizing the high-entropy tokens that decide reasoning directions. Collectively, our results highlight the potential to understand RLVR through a token-entropy perspective and optimize RLVR by leveraging high-entropy minority tokens to further improve LLM reasoning.
- [401] arXiv:2506.01945 (cross-list from econ.EM) [pdf, html, other]
-
Title: Stock Market Telepathy: Graph Neural Networks Predicting the Secret Conversations between MINT and G7 CountriesSubjects: Econometrics (econ.EM); Machine Learning (cs.LG); Applications (stat.AP)
Emerging economies, particularly the MINT countries (Mexico, Indonesia, Nigeria, and Türkiye), are gaining influence in global stock markets, although they remain susceptible to the economic conditions of developed countries like the G7 (Canada, France, Germany, Italy, Japan, the United Kingdom, and the United States). This interconnectedness and sensitivity of financial markets make understanding these relationships crucial for investors and policymakers to predict stock price movements accurately. To this end, we examined the main stock market indices of G7 and MINT countries from 2012 to 2024, using a recent graph neural network (GNN) algorithm called multivariate time series forecasting with graph neural network (MTGNN). This method allows for considering complex spatio-temporal connections in multivariate time series. In the implementations, MTGNN revealed that the US and Canada are the most influential G7 countries regarding stock indices in the forecasting process, and Indonesia and Türkiye are the most influential MINT countries. Additionally, our results showed that MTGNN outperformed traditional methods in forecasting the prices of stock market indices for MINT and G7 countries. Consequently, the study offers valuable insights into economic blocks' markets and presents a compelling empirical approach to analyzing global stock market dynamics using MTGNN.
- [402] arXiv:2506.01951 (cross-list from cs.CL) [pdf, html, other]
-
Title: Self-ensemble: Mitigating Confidence Distortion for Large Language ModelsZicheng Xu, Guanchu Wang, Guangyao Zheng, Yu-Neng Chuang, Alexander Szalay, Xia Hu, Vladimir BravermanSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Although Large Language Models (LLMs) perform well in general fields, they exhibit a confidence distortion problem on multi-choice question-answering (MCQA), particularly as the number of answer choices increases. Specifically, on MCQA with many choices, LLMs suffer from under-confidence in correct predictions and over-confidence in incorrect ones, leading to a substantially degraded performance. To solve this problem, we propose Self-ensemble in this work. Our method splits the choices into several groups and ensembles LLM predictions across these groups to reach a final decision. The advantage of Self-ensemble is its plug-and-play nature, where it can be integrated into existing LLM architecture based on a designed attention mask and positional encoding, without requiring labeled datasets for parameter tuning. Experimental results on three LLMs and datasets demonstrate that Self-ensemble comprehensively addresses the confidence distortion problem of LLMs, outperforming standard inference as well as baseline methods.
- [403] arXiv:2506.01952 (cross-list from cs.CL) [pdf, other]
-
Title: WebChoreArena: Evaluating Web Browsing Agents on Realistic Tedious Web TasksAtsuyuki Miyai, Zaiying Zhao, Kazuki Egashira, Atsuki Sato, Tatsumi Sunada, Shota Onohara, Hiromasa Yamanishi, Mashiro Toyooka, Kunato Nishina, Ryoma Maeda, Kiyoharu Aizawa, Toshihiko YamasakiComments: Project Page: this https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Powered by a large language model (LLM), a web browsing agent operates web browsers in a human-like manner and offers a highly transparent path toward automating a wide range of everyday tasks. As web agents become increasingly capable and demonstrate proficiency in general browsing tasks, a critical question emerges: Can they go beyond general browsing to robustly handle tasks that are tedious and complex, or chores that humans often avoid doing themselves? In this paper, we introduce WebChoreArena, a new fully reproducible benchmark comprising 532 carefully curated tasks designed to extend the scope of WebArena beyond general browsing to more labor-intensive and tedious tasks. WebChoreArena systematically integrates three key challenges: (i) Massive Memory tasks requiring accurate retrieval of large amounts of information in the observations, (ii) Calculation tasks demanding precise mathematical reasoning, and (iii) Long-Term Memory tasks necessitating long-term memory across multiple webpages. Built on top of the fully reproducible and widely adopted four WebArena simulation environments, WebChoreArena ensures strict reproducibility and enables fair, direct comparisons with the established WebArena benchmark, offering key insights into agent progress. Our experimental results demonstrate that as LLMs evolve, represented by GPT-4o, Claude 3.7 Sonnet, and Gemini 2.5 Pro, significant improvements in performance are observed on WebChoreArena. These findings suggest that WebChoreArena is well-suited to measure the advancement of state-of-the-art LLMs with greater clarity. Nevertheless, the results also indicate that even with Gemini 2.5 Pro, there remains substantial room for improvement compared to WebArena, highlighting the increased challenges posed by WebChoreArena.
- [404] arXiv:2506.01954 (cross-list from cs.CL) [pdf, html, other]
-
Title: DRAG: Distilling RAG for SLMs from LLMs to Transfer Knowledge and Mitigate Hallucination via Evidence and Graph-based DistillationJennifer Chen, Aidar Myrzakhan, Yaxin Luo, Hassaan Muhammad Khan, Sondos Mahmoud Bsharat, Zhiqiang ShenComments: ACL 2025 Main. Code is available at this https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Retrieval-Augmented Generation (RAG) methods have proven highly effective for tasks requiring factual consistency and robust knowledge retrieval. However, large-scale RAG systems consume significant computational resources and are prone to generating hallucinated content from Humans. In this work, we introduce $\texttt{DRAG}$, a novel framework for distilling RAG knowledge from large-scale Language Models (LLMs) into small LMs (SLMs). Our approach leverages evidence- and knowledge graph-based distillation, ensuring that the distilled model retains critical factual knowledge while significantly reducing model size and computational cost. By aligning the smaller model's predictions with a structured knowledge graph and ranked evidence, $\texttt{DRAG}$ effectively mitigates hallucinations and improves factual accuracy. We further present a case demonstrating how our framework mitigates user privacy risks and introduce a corresponding benchmark. Experimental evaluations on multiple benchmarks demonstrate that our method outperforms the prior competitive RAG methods like MiniRAG for SLMs by up to 27.7% using the same models, preserving high-level efficiency and reliability. With $\texttt{DRAG}$, we provide a practical and resource-efficient roadmap to deploying enhanced retrieval and generation capabilities in small-sized LLMs.
- [405] arXiv:2506.01955 (cross-list from cs.CV) [pdf, html, other]
-
Title: Dual-Process Image GenerationSubjects: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL); Machine Learning (cs.LG)
Prior methods for controlling image generation are limited in their ability to be taught new tasks. In contrast, vision-language models, or VLMs, can learn tasks in-context and produce the correct outputs for a given input. We propose a dual-process distillation scheme that allows feed-forward image generators to learn new tasks from deliberative VLMs. Our scheme uses a VLM to rate the generated images and backpropagates this gradient to update the weights of the image generator. Our general framework enables a wide variety of new control tasks through the same text-and-image based interface. We showcase a handful of applications of this technique for different types of control signals, such as commonsense inferences and visual prompts. With our method, users can implement multimodal controls for properties such as color palette, line weight, horizon position, and relative depth within a matter of minutes. Project page: this https URL.
Cross submissions (showing 167 of 167 entries)
- [406] arXiv:2205.09337 (replaced) [pdf, other]
-
Title: Deep Learning in Business Analytics: A Clash of Expectations and RealityJournal-ref: International Journal of Information Management Data Insights, Volume 3, Issue 1, 2023, 100146, ISSN 2667-0968Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Databases (cs.DB); Risk Management (q-fin.RM)
Our fast-paced digital economy shaped by global competition requires increased data-driven decision-making based on artificial intelligence (AI) and machine learning (ML). The benefits of deep learning (DL) are manifold, but it comes with limitations that have, so far, interfered with widespread industry adoption. This paper explains why DL, despite its popularity, has difficulties speeding up its adoption within business analytics. It is shown that the adoption of deep learning is not only affected by computational complexity, lacking big data architecture, lack of transparency (black-box), skill shortage, and leadership commitment, but also by the fact that DL does not outperform traditional ML models in the case of structured datasets with fixed-length feature vectors. Deep learning should be regarded as a powerful addition to the existing body of ML models instead of a one size fits all solution. The results strongly suggest that gradient boosting can be seen as the go-to model for predictions on structured datasets within business analytics. In addition to the empirical study based on three industry use cases, the paper offers a comprehensive discussion of those results, practical implications, and a roadmap for future research.
- [407] arXiv:2205.10538 (replaced) [pdf, other]
-
Title: Automated machine learning: AI-driven decision making in business analyticsJournal-ref: Intelligent Systems with Applications, Volume 18, 2023, 200188, ISSN 2667-3053Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
The realization that AI-driven decision-making is indispensable in today's fast-paced and ultra-competitive marketplace has raised interest in industrial machine learning (ML) applications significantly. The current demand for analytics experts vastly exceeds the supply. One solution to this problem is to increase the user-friendliness of ML frameworks to make them more accessible for the non-expert. Automated machine learning (AutoML) is an attempt to solve the problem of expertise by providing fully automated off-the-shelf solutions for model choice and hyperparameter tuning. This paper analyzed the potential of AutoML for applications within business analytics, which could help to increase the adoption rate of ML across all industries. The H2O AutoML framework was benchmarked against a manually tuned stacked ML model on three real-world datasets. The manually tuned ML model could reach a performance advantage in all three case studies used in the experiment. Nevertheless, the H2O AutoML package proved to be quite potent. It is fast, easy to use, and delivers reliable results, which come close to a professionally tuned ML model. The H2O AutoML framework in its current capacity is a valuable tool to support fast prototyping with the potential to shorten development and deployment cycles. It can also bridge the existing gap between supply and demand for ML experts and is a big step towards automated decisions in business analytics. Finally, AutoML has the potential to foster human empowerment in a world that is rapidly becoming more automated and digital.
- [408] arXiv:2209.01847 (replaced) [pdf, html, other]
-
Title: Conflict-Aware Pseudo Labeling via Optimal Transport for Entity AlignmentComments: Accepted by ICDM 2022Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Entity alignment aims to discover unique equivalent entity pairs with the same meaning across different knowledge graphs (KGs). Existing models have focused on projecting KGs into a latent embedding space so that inherent semantics between entities can be captured for entity alignment. However, the adverse impacts of alignment conflicts have been largely overlooked during training, thereby limiting the entity alignment performance. To address this issue, we propose a novel Conflict-aware Pseudo Labeling via Optimal Transport model (CPL-OT) for entity alignment. The key idea is to iteratively pseudo-label alignment pairs empowered with conflict-aware optimal transport (OT) modeling to boost the precision of entity alignment. CPL-OT is composed of two key components -- entity embedding learning with global-local aggregation and iterative conflict-aware pseudo labeling -- that mutually reinforce each other. To mitigate alignment conflicts during pseudo labeling, we propose to use optimal transport as an effective means to warrant one-to-one entity alignment between two KGs with the minimal overall transport cost. Extensive experiments on benchmark datasets validate the superiority of CPL-OT over state-of-the-art baselines under both settings with and without prior alignment seeds.
- [409] arXiv:2303.17475 (replaced) [pdf, html, other]
-
Title: Learning distributed representations with efficient SoftMax normalizationJournal-ref: Transactions on Machine Learning Research, 2835-8856, 2025Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL); Machine Learning (stat.ML)
Learning distributed representations, or embeddings, that encode the relational similarity patterns among objects is a relevant task in machine learning. A popular method to learn the embedding matrices $X, Y$ is optimizing a loss function of the term ${\rm SoftMax}(XY^T)$. The complexity required to calculate this term, however, runs quadratically with the problem size, making it a computationally heavy solution. In this article, we propose a linear-time heuristic approximation to compute the normalization constants of ${\rm SoftMax}(XY^T)$ for embedding vectors with bounded norms. We show on some pre-trained embedding datasets that the proposed estimation method achieves higher or comparable accuracy with competing methods. From this result, we design an efficient and task-agnostic algorithm that learns the embeddings by optimizing the cross entropy between the softmax and a set of probability distributions given as inputs. The proposed algorithm is interpretable and easily adapted to arbitrary embedding problems. We consider a few use cases and observe similar or higher performances and a lower computational time than similar ``2Vec'' algorithms.
- [410] arXiv:2304.09276 (replaced) [pdf, html, other]
-
Title: Towards a Neural Lambda Calculus: Neurosymbolic AI Applied to the Foundations of Functional ProgrammingComments: Keywords: Machine Learning, Lambda Calculus, Neurosymbolic AI, Neural Networks, Transformer Model, Sequence-to-Sequence Models, Computational ModelsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Logic in Computer Science (cs.LO)
Over the last decades, deep neural networks based-models became the dominant paradigm in machine learning. Further, the use of artificial neural networks in symbolic learning has been seen as increasingly relevant recently. To study the capabilities of neural networks in the symbolic AI domain, researchers have explored the ability of deep neural networks to learn mathematical constructions, such as addition and multiplication, logic inference, such as theorem provers, and even the execution of computer programs. The latter is known to be too complex a task for neural networks. Therefore, the results were not always successful, and often required the introduction of biased elements in the learning process, in addition to restricting the scope of possible programs to be executed. In this work, we will analyze the ability of neural networks to learn how to execute programs as a whole. To do so, we propose a different approach. Instead of using an imperative programming language, with complex structures, we use the Lambda Calculus ({\lambda}-Calculus), a simple, but Turing-Complete mathematical formalism, which serves as the basis for modern functional programming languages and is at the heart of computability theory. We will introduce the use of integrated neural learning and lambda calculi formalization. Finally, we explore execution of a program in {\lambda}-Calculus is based on reductions, we will show that it is enough to learn how to perform these reductions so that we can execute any program. Keywords: Machine Learning, Lambda Calculus, Neurosymbolic AI, Neural Networks, Transformer Model, Sequence-to-Sequence Models, Computational Models
- [411] arXiv:2304.11265 (replaced) [pdf, other]
-
Title: Estimating Motor Symptom Presence and Severity in Parkinson's Disease from Wrist Accelerometer Time Series using ROCKET and InceptionTimeComments: The source code is available from this https URLJournal-ref: Scientific Reports. 31 May 2025. Vol. 15, no. 1, p. 19140. https://www.nature.com/articles/s41598-025-04263-2Subjects: Machine Learning (cs.LG)
Parkinson's disease (PD) is a neurodegenerative condition characterized by frequently changing motor symptoms, necessitating continuous symptom monitoring for more targeted treatment. Classical time series classification and deep learning techniques have demonstrated limited efficacy in monitoring PD symptoms using wearable accelerometer data due to complex PD movement patterns and the small size of available datasets. We investigate InceptionTime and RandOm Convolutional KErnel Transform (ROCKET) as they are promising for PD symptom monitoring. InceptionTime's high learning capacity is well-suited to modeling complex movement patterns, while ROCKET is suited to small datasets. With random search methodology, we identify the highest-scoring InceptionTime architecture and compare its performance to ROCKET with a ridge classifier and a multi-layer perceptron (MLP) on wrist motion data from PD patients. Our findings indicate that all approaches can learn to estimate tremor severity and bradykinesia presence with moderate performance but encounter challenges in detecting dyskinesia. Among the presented approaches, ROCKET demonstrates higher scores in identifying dyskinesia, whereas InceptionTime exhibits slightly better performance in tremor and bradykinesia estimation. Notably, both methods outperform the multi-layer perceptron. In conclusion, InceptionTime can classify complex wrist motion time series and holds potential for continuous symptom monitoring in PD with further development.
- [412] arXiv:2309.12701 (replaced) [pdf, html, other]
-
Title: Breiman meets Bellman: Non-Greedy Decision Trees with MDPsComments: ACM SIGKDD 2025, 12 pagesSubjects: Machine Learning (cs.LG)
In supervised learning, decision trees are valued for their interpretability and performance. While greedy decision tree algorithms like CART remain widely used due to their computational efficiency, they often produce sub-optimal solutions with respect to a regularized training loss. Conversely, optimal decision tree methods can find better solutions but are computationally intensive and typically limited to shallow trees or binary features. We present Dynamic Programming Decision Trees (DPDT), a framework that bridges the gap between greedy and optimal approaches. DPDT relies on a Markov Decision Process formulation combined with heuristic split generation to construct near-optimal decision trees with significantly reduced computational complexity. Our approach dynamically limits the set of admissible splits at each node while directly optimizing the tree regularized training loss. Theoretical analysis demonstrates that DPDT can minimize regularized training losses at least as well as CART. Our empirical study shows on multiple datasets that DPDT achieves near-optimal loss with orders of magnitude fewer operations than existing optimal solvers. More importantly, extensive benchmarking suggests statistically significant improvements of DPDT over both CART and optimal decision trees in terms of generalization to unseen data. We demonstrate DPDT practicality through applications to boosting, where it consistently outperforms baselines. Our framework provides a promising direction for developing efficient, near-optimal decision tree algorithms that scale to practical applications.
- [413] arXiv:2310.06417 (replaced) [pdf, html, other]
-
Title: Supercharging Graph Transformers with Advective DiffusionComments: Accepted to ICML 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
The capability of generalization is a cornerstone for the success of modern learning systems. For non-Euclidean data, e.g., graphs, that particularly involves topological structures, one important aspect neglected by prior studies is how machine learning models generalize under topological shifts. This paper proposes AdvDIFFormer, a physics-inspired graph Transformer model designed to address this challenge. The model is derived from advective diffusion equations which describe a class of continuous message passing process with observed and latent topological structures. We show that AdvDIFFormer has provable capability for controlling generalization error with topological shifts, which in contrast cannot be guaranteed by graph diffusion models. Empirically, the model demonstrates superiority in various predictive tasks across information networks, molecular screening and protein interactions.
- [414] arXiv:2310.13391 (replaced) [pdf, html, other]
-
Title: Learning Successor Features with Distributed Hebbian Temporal MemoryComments: Poster on ICLR 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE)
This paper presents a novel approach to address the challenge of online sequence learning for decision making under uncertainty in non-stationary, partially observable environments. The proposed algorithm, Distributed Hebbian Temporal Memory (DHTM), is based on the factor graph formalism and a multi-component neuron model. DHTM aims to capture sequential data relationships and make cumulative predictions about future observations, forming Successor Features (SFs). Inspired by neurophysiological models of the neocortex, the algorithm uses distributed representations, sparse transition matrices, and local Hebbian-like learning rules to overcome the instability and slow learning of traditional temporal memory algorithms such as RNN and HMM. Experimental results show that DHTM outperforms LSTM, RWKV and a biologically inspired HMM-like algorithm, CSCG, on non-stationary data sets. Our results suggest that DHTM is a promising approach to address the challenges of online sequence learning and planning in dynamic environments.
- [415] arXiv:2310.20363 (replaced) [pdf, other]
-
Title: Hidden Conflicts in Neural Networks and Their Implications for ExplainabilityComments: Accepted at FAccT 2025Subjects: Machine Learning (cs.LG)
Artificial Neural Networks (ANNs) often represent conflicts between features, arising naturally during training as the network learns to integrate diverse and potentially disagreeing inputs to better predict the target variable. Despite their relevance to the ``reasoning'' processes of these models, the properties and implications of conflicts for understanding and explaining ANNs remain underexplored. In this paper, we develop a rigorous theory of conflicts in ANNs and demonstrate their impact on ANN explainability through two case studies. In the first case study, we use our theory of conflicts to inspire the design of a novel feature attribution method, which we call Conflict-Aware Feature-wise Explanations (CAFE). CAFE separates the positive and negative influences of features and biases, enabling more faithful explanations for models applied to tabular data. In the second case study, we take preliminary steps towards understanding the role of conflicts in out-of-distribution (OOD) scenarios. Through our experiments, we identify potentially useful connections between model conflicts and different kinds of distributional shifts in tabular and image data. Overall, our findings demonstrate the importance of accounting for conflicts in the development of more reliable explanation methods for AI systems, which are crucial for the beneficial use of these systems in the society.
- [416] arXiv:2311.16646 (replaced) [pdf, html, other]
-
Title: Rethinking Backdoor Attacks on Dataset Distillation: A Kernel Method PerspectiveComments: 21 pages, 1 figures, 17 tablesJournal-ref: Proceedings of the Twelfth International Conference on Learning Representations (ICLR 2024)Subjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
Dataset distillation offers a potential means to enhance data efficiency in deep learning. Recent studies have shown its ability to counteract backdoor risks present in original training samples. In this study, we delve into the theoretical aspects of backdoor attacks and dataset distillation based on kernel methods. We introduce two new theory-driven trigger pattern generation methods specialized for dataset distillation. Following a comprehensive set of analyses and experiments, we show that our optimization-based trigger design framework informs effective backdoor attacks on dataset distillation. Notably, datasets poisoned by our designed trigger prove resilient against conventional backdoor attack detection and mitigation methods. Our empirical results validate that the triggers developed using our approaches are proficient at executing resilient backdoor attacks.
- [417] arXiv:2401.06279 (replaced) [pdf, other]
-
Title: Sampling and Uniqueness Sets in Graphon Signal ProcessingSubjects: Machine Learning (cs.LG); Signal Processing (eess.SP)
In this work, we study the properties of sampling sets on families of large graphs by leveraging the theory of graphons and graph limits. To this end, we extend to graphon signals the notion of removable and uniqueness sets, which was developed originally for the analysis of signals on graphs. We state the formal definition of a $\Lambda-$removable set and conditions under which a bandlimited graphon signal can be represented in a unique way when its samples are obtained from the complement of a given $\Lambda-$removable set in the graphon. By leveraging such results we show that graphon representations of graphs and graph signals can be used as a common framework to compare sampling sets between graphs with different numbers of nodes and edges, and different node labelings. Additionally, given a sequence of graphs that converges to a graphon, we show that the sequences of sampling sets whose graphon representation is identical in $[0,1]$ are convergent as well. We exploit the convergence results to provide an algorithm that obtains approximately close to optimal sampling sets. Performing a set of numerical experiments, we evaluate the quality of these sampling sets. Our results open the door for the efficient computation of optimal sampling sets in graphs of large size.
- [418] arXiv:2401.06634 (replaced) [pdf, html, other]
-
Title: CCFC: Bridging Federated Clustering and Contrastive LearningSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Federated clustering, an essential extension of centralized clustering for federated scenarios, enables multiple data-holding clients to collaboratively group data while keeping their data locally. In centralized scenarios, clustering driven by representation learning has made significant advancements in handling high-dimensional complex data. However, the combination of federated clustering and representation learning remains underexplored. To bridge this, we first tailor a cluster-contrastive model for learning clustering-friendly representations. Then, we harness this model as the foundation for proposing a new federated clustering method, named cluster-contrastive federated clustering (CCFC). Benefiting from representation learning, the clustering performance of CCFC even double those of the best baseline methods in some cases. Compared to the most related baseline, the benefit results in substantial NMI score improvements of up to 0.4155 on the most conspicuous case. Moreover, CCFC also shows superior performance in handling device failures from a practical viewpoint.
- [419] arXiv:2401.13796 (replaced) [pdf, other]
-
Title: Don't Push the Button! Exploring Data Leakage Risks in Machine Learning and Transfer LearningComments: under reviewSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Machine Learning (ML) has revolutionized various domains, offering predictive capabilities in several areas. However, with the increasing accessibility of ML tools, many practitioners, lacking deep ML expertise, adopt a "push the button" approach, utilizing user-friendly interfaces without a thorough understanding of underlying algorithms. While this approach provides convenience, it raises concerns about the reliability of outcomes, leading to challenges such as incorrect performance evaluation. This paper addresses a critical issue in ML, known as data leakage, where unintended information contaminates the training data, impacting model performance evaluation. Users, due to a lack of understanding, may inadvertently overlook crucial steps, leading to optimistic performance estimates that may not hold in real-world scenarios. The discrepancy between evaluated and actual performance on new data is a significant concern. In particular, this paper categorizes data leakage in ML, discussing how certain conditions can propagate through the ML workflow. Furthermore, it explores the connection between data leakage and the specific task being addressed, investigates its occurrence in Transfer Learning, and compares standard inductive ML with transductive ML frameworks. The conclusion summarizes key findings, emphasizing the importance of addressing data leakage for robust and reliable ML applications.
- [420] arXiv:2401.17539 (replaced) [pdf, html, other]
-
Title: Gradient-Free Score-Based Sampling Methods with EnsemblesJournal-ref: Applied Mathematical Modelling (2025), 147,Subjects: Machine Learning (cs.LG); Computation (stat.CO)
Recent developments in generative modeling have utilized score-based methods coupled with stochastic differential equations to sample from complex probability distributions. However, these and other performant sampling methods generally require gradients of the target probability distribution, which can be unavailable or computationally prohibitive in many scientific and engineering applications. Here, we introduce ensembles within score-based sampling methods to develop gradient-free approximate sampling techniques that leverage the collective dynamics of particle ensembles to compute approximate reverse diffusion drifts. We introduce the underlying methodology, emphasizing its relationship with generative diffusion models and the previously introduced Föllmer sampler. We demonstrate the efficacy of the ensemble strategies through various examples, ranging from low- to medium-dimensionality sampling problems, including multi-modal and highly non-Gaussian probability distributions, and provide comparisons to traditional methods like the No-U-Turn Sampler. Additionally, we showcase these strategies in the context of a high-dimensional Bayesian inversion problem within the geophysical sciences. Our findings highlight the potential of ensemble strategies for modeling complex probability distributions in situations where gradients are unavailable.
- [421] arXiv:2402.04435 (replaced) [pdf, html, other]
-
Title: PreGIP: Watermarking the Pretraining of Graph Neural Networks for Deep Intellectual Property ProtectionSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Pretraining on Graph Neural Networks (GNNs) has shown great power in facilitating various downstream tasks. As pretraining generally requires huge amount of data and computational resources, the pretrained GNNs are high-value Intellectual Properties (IP) of the legitimate owner. However, adversaries may illegally copy and deploy the pretrained GNN models for their downstream tasks. Though initial efforts have been made to watermark GNN classifiers for IP protection, these methods require the target classification task for watermarking, and thus are not applicable to self-supervised pretraining of GNN models. Hence, in this work, we propose a novel framework named PreGIP to watermark the pretraining of GNN encoder for IP protection while maintain the high-quality of the embedding space. PreGIP incorporates a task-free watermarking loss to watermark the embedding space of pretrained GNN encoder. A finetuning-resistant watermark injection is further deployed. Theoretical analysis and extensive experiments show the effectiveness of {\method} in IP protection and maintaining high-performance for downstream tasks.
- [422] arXiv:2402.05806 (replaced) [pdf, html, other]
-
Title: On Temperature Scaling and Conformal Prediction of Deep ClassifiersComments: ICML 2025 (camera-ready). Code can be found at: this https URLSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
In many classification applications, the prediction of a deep neural network (DNN) based classifier needs to be accompanied by some confidence indication. Two popular approaches for that aim are: 1) Calibration: modifies the classifier's softmax values such that the maximal value better estimates the correctness probability; and 2) Conformal Prediction (CP): produces a prediction set of candidate labels that contains the true label with a user-specified probability, guaranteeing marginal coverage but not, e.g., per class coverage. In practice, both types of indications are desirable, yet, so far the interplay between them has not been investigated. Focusing on the ubiquitous Temperature Scaling (TS) calibration, we start this paper with an extensive empirical study of its effect on prominent CP methods. We show that while TS calibration improves the class-conditional coverage of adaptive CP methods, surprisingly, it negatively affects their prediction set sizes. Motivated by this behavior, we explore the effect of TS on CP beyond its calibration application and reveal an intriguing trend under which it allows to trade prediction set size and conditional coverage of adaptive CP methods. Then, we establish a mathematical theory that explains the entire non-monotonic trend. Finally, based on our experiments and theory, we offer simple guidelines for practitioners to effectively combine adaptive CP with calibration, aligned with user-defined goals.
- [423] arXiv:2402.15734 (replaced) [pdf, html, other]
-
Title: Data-Efficient Operator Learning via Unsupervised Pretraining and In-Context LearningSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Recent years have witnessed the promise of coupling machine learning methods and physical domain-specific insights for solving scientific problems based on partial differential equations (PDEs). However, being data-intensive, these methods still require a large amount of PDE data. This reintroduces the need for expensive numerical PDE solutions, partially undermining the original goal of avoiding these expensive simulations. In this work, seeking data efficiency, we design unsupervised pretraining for PDE operator learning. To reduce the need for training data with heavy simulation costs, we mine unlabeled PDE data without simulated solutions, and we pretrain neural operators with physics-inspired reconstruction-based proxy tasks. To improve out-of-distribution performance, we further assist neural operators in flexibly leveraging a similarity-based method that learns in-context examples, without incurring extra training costs or designs. Extensive empirical evaluations on a diverse set of PDEs demonstrate that our method is highly data-efficient, more generalizable, and even outperforms conventional vision-pretrained models. We provide our code at this https URL.
- [424] arXiv:2403.08291 (replaced) [pdf, html, other]
-
Title: CleanAgent: Automating Data Standardization with LLM-based AgentsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
Data standardization is a crucial part of the data science life cycle. While tools like Pandas offer robust functionalities, their complexity and the manual effort required for customizing code to diverse column types pose significant challenges. Although large language models (LLMs) like ChatGPT have shown promise in automating this process through natural language understanding and code generation, it still demands expert-level programming knowledge and continuous interaction for prompt refinement. To solve these challenges, our key idea is to propose a Python library with declarative, unified APIs for standardizing different column types, simplifying the LLM's code generation with concise API calls. We first propose this http URL, a component of the Dataprep Python Library, significantly reduces the coding complexity by enabling the standardization of specific column types with a single line of code. Then, we introduce the CleanAgent framework integrating this http URL and LLM-based agents to automate the data standardization process. With CleanAgent, data scientists only need to provide their requirements once, allowing for a hands-free process. To demonstrate the practical utility of CleanAgent, we developed a user-friendly web application, allowing users to interact with it using real-world datasets.
- [425] arXiv:2404.02926 (replaced) [pdf, html, other]
-
Title: Log-PDE Methods for Rough Signature KernelsSubjects: Machine Learning (cs.LG); Analysis of PDEs (math.AP)
Signature kernels, inner products of path signatures, underpin several machine learning algorithms for multivariate time series analysis. For bounded variation paths, signature kernels were recently shown to solve a Goursat PDE. However, existing PDE solvers only use increments as input data, leading to first order approximation errors. These approaches become computationally intractable for highly oscillatory input paths, as they have to be resolved at a fine enough scale to accurately recover their signature kernel, resulting in significant time and memory complexities. In this paper, we extend the analysis to rough paths, and show, leveraging the framework of smooth rough paths, that the resulting rough signature kernels can be approximated by a novel system of PDEs whose coefficients involve higher order iterated integrals of the input rough paths. We show that this system of PDEs admits a unique solution and establish quantitative error bounds yielding a higher order approximation to rough signature kernels.
- [426] arXiv:2404.10512 (replaced) [pdf, other]
-
Title: Four-hour thunderstorm nowcasting using deep diffusion models of satelliteSubjects: Machine Learning (cs.LG)
Convection (thunderstorm) develops rapidly within hours and is highly destructive, posing a significant challenge for nowcasting and resulting in substantial losses to nature and society. After the emergence of artificial intelligence (AI)-based methods, convection nowcasting has experienced rapid advancements, with its performance surpassing that of physics-based numerical weather prediction and other conventional approaches. However, the lead time and coverage of it still leave much to be desired and hardly meet the needs of disaster emergency response. Here, we propose deep diffusion models of satellite (DDMS) to establish an AI-based convection nowcasting system. Specifically, DDMS employs diffusion processes to effectively simulate complicated spatiotemporal evolution patterns of convective clouds, significantly improving the forecast lead time. Additionally, it combines geostationary satellite brightness temperature data and domain knowledge from meteorological experts, thereby achieving planetary-scale forecast coverage. During long-term tests and objective validation based on the FengYun-4A satellite, our system achieves, for the first time, effective convection nowcasting up to 4 hours, with broad coverage (about 20,000,000 km$^2$), remarkable accuracy, and high resolution (15 minutes; 4 km). Its performance reaches a new height in convection nowcasting compared to the existing models. In terms of application, our system is highly transferable with the potential to collaborate with multiple satellites for global convection nowcasting. Furthermore, our results highlight the remarkable capabilities of diffusion models in convective clouds forecasting, as well as the significant value of geostationary satellite data when empowered by AI technologies.
- [427] arXiv:2404.14202 (replaced) [pdf, html, other]
-
Title: An Adaptive Approach for Infinitely Many-armed Bandits under Generalized Rotting ConstraintsComments: NeurIPS 2024Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
In this study, we consider the infinitely many-armed bandit problems in a rested rotting setting, where the mean reward of an arm may decrease with each pull, while otherwise, it remains unchanged. We explore two scenarios regarding the rotting of rewards: one in which the cumulative amount of rotting is bounded by $V_T$, referred to as the slow-rotting case, and the other in which the cumulative number of rotting instances is bounded by $S_T$, referred to as the abrupt-rotting case. To address the challenge posed by rotting rewards, we introduce an algorithm that utilizes UCB with an adaptive sliding window, designed to manage the bias and variance trade-off arising due to rotting rewards. Our proposed algorithm achieves tight regret bounds for both slow and abrupt rotting scenarios. Lastly, we demonstrate the performance of our algorithm using numerical experiments.
- [428] arXiv:2405.00172 (replaced) [pdf, html, other]
-
Title: Bypassing Skip-Gram Negative Sampling: Dimension Regularization as a More Efficient Alternative for Graph EmbeddingsComments: Published in KDD'25Subjects: Machine Learning (cs.LG); Social and Information Networks (cs.SI); Machine Learning (stat.ML)
A wide range of graph embedding objectives decompose into two components: one that enforces similarity, attracting the embeddings of nodes that are perceived as similar, and another that enforces dissimilarity, repelling the embeddings of nodes that are perceived as dissimilar. Without repulsion, the embeddings would collapse into trivial solutions. Skip-Gram Negative Sampling (SGNS) is a popular and efficient repulsion approach that prevents collapse by repelling each node from a sample of dissimilar nodes. In this work, we show that when repulsion is most needed and the embeddings approach collapse, SGNS node-wise repulsion is, in the aggregate, an approximate re-centering of the node embedding dimensions. Such dimension operations are more scalable than node operations and produce a simpler geometric interpretation of the repulsion. Our theoretical result establishes dimension regularization as an effective and more efficient, compared to skip-gram node contrast, approach to enforcing dissimilarity among embeddings of nodes. We use this result to propose a flexible algorithm augmentation framework that improves the scalability of any existing algorithm using SGNS. The framework prioritizes node attraction and replaces SGNS with dimension regularization. We instantiate this generic framework for LINE and node2vec and show that the augmented algorithms preserve downstream link-prediction performance while reducing GPU memory usage by up to 33.3% and training time by 23.4%. Moreover, we show that completely removing repulsion (a special case of our augmentation framework) in LINE reduces training time by 70.9% on average, while increasing link prediction performance, especially for graphs that are globally sparse but locally dense. In general, however, repulsion is needed, and dimension regularization provides an efficient alternative to SGNS.
- [429] arXiv:2405.13698 (replaced) [pdf, html, other]
-
Title: How to set AdamW's weight decay as you scale model and dataset sizeComments: Published in ICML 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
The scaling of the optimal AdamW weight decay hyperparameter with model and dataset size is critical as we seek to build larger models, but is poorly understood. We show that weights learned by AdamW can be understood as an exponential moving average (EMA) of recent updates. This gives critical insights for how to set the weight decay in AdamW, and how the weight decay should scale with model and dataset size. In particular, the key hyperparameter for an exponential moving average is the EMA timescale. Intuitively, the EMA timescale can be understood as the number of recent iterations the EMA averages over. We find that the optimal timescale, measured in epochs, is roughly constant as we change model and dataset size. Moreover, given a learning rate, there is a one-to-one mapping from the EMA timescale to the weight decay hyperparameter. Thus, if the optimal EMA timescale is constant, that implies that as the dataset size increases, the optimal weight decay should fall and as the model size increases, the optimal weight decay should increase (if we follow the muP recommendation for scaling the learning rate). We validate these scaling rules on ResNet-18 and Vision Transformers trained on CIFAR-10 and ImageNet, and on NanoGPT pre-training on OpenWebText. Finally, we found that as training progresses, muP's learning rate scaling breaks down for AdamW unless weight decay is scaled appropriately.
- [430] arXiv:2405.16255 (replaced) [pdf, html, other]
-
Title: GeoAdaLer: Geometric Insights into Adaptive Stochastic Gradient Descent AlgorithmsSubjects: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
The Adam optimization method has achieved remarkable success in addressing contemporary challenges in stochastic optimization. This method falls within the realm of adaptive sub-gradient techniques, yet the underlying geometric principles guiding its performance have remained shrouded in mystery, and have long confounded researchers. In this paper, we introduce GeoAdaLer (Geometric Adaptive Learner), a novel adaptive learning method for stochastic gradient descent optimization, which draws from the geometric properties of the optimization landscape. Beyond emerging as a formidable contender, the proposed method extends the concept of adaptive learning by introducing a geometrically inclined approach that enhances the interpretability and effectiveness in complex optimization scenarios
- [431] arXiv:2405.16828 (replaced) [pdf, html, other]
-
Title: Kernel-based Optimally Weighted Conformal Prediction IntervalsSubjects: Machine Learning (cs.LG); Statistics Theory (math.ST); Machine Learning (stat.ML)
In this work, we present a novel conformal prediction method for time-series, which we call Kernel-based Optimally Weighted Conformal Prediction Intervals (KOWCPI). Specifically, KOWCPI adapts the classic Reweighted Nadaraya-Watson (RNW) estimator for quantile regression on dependent data and learns optimal data-adaptive weights. Theoretically, we tackle the challenge of establishing a conditional coverage guarantee for non-exchangeable data under strong mixing conditions on the non-conformity scores. We demonstrate the superior performance of KOWCPI on real and synthetic time-series data against state-of-the-art methods, where KOWCPI achieves narrower confidence intervals without losing coverage.
- [432] arXiv:2405.17829 (replaced) [pdf, html, other]
-
Title: LDMol: A Text-to-Molecule Diffusion Model with Structurally Informative Latent Space Surpasses AR ModelsComments: Poster in ICML 2025; 19 pages, 13 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
With the emergence of diffusion models as a frontline generative model, many researchers have proposed molecule generation techniques with conditional diffusion models. However, the unavoidable discreteness of a molecule makes it difficult for a diffusion model to connect raw data with highly complex conditions like natural language. To address this, here we present a novel latent diffusion model dubbed LDMol for text-conditioned molecule generation. By recognizing that the suitable latent space design is the key to the diffusion model performance, we employ a contrastive learning strategy to extract novel feature space from text data that embeds the unique characteristics of the molecule structure. Experiments show that LDMol outperforms the existing autoregressive baselines on the text-to-molecule generation benchmark, being one of the first diffusion models that outperforms autoregressive models in textual data generation with a better choice of the latent domain. Furthermore, we show that LDMol can be applied to downstream tasks such as molecule-to-text retrieval and text-guided molecule editing, demonstrating its versatility as a diffusion model.
- [433] arXiv:2405.18311 (replaced) [pdf, html, other]
-
Title: Deterministic and statistical calibration of constitutive models from full-field data with parametric physics-informed neural networksDavid Anton, Jendrik-Alexander Tröger, Henning Wessels, Ulrich Römer, Alexander Henkes, Stefan HartmannSubjects: Machine Learning (cs.LG)
The calibration of constitutive models from full-field data has recently gained increasing interest due to improvements in full-field measurement capabilities. In addition to the experimental characterization of novel materials, continuous structural health monitoring is another application that is of great interest. However, monitoring is usually associated with severe time constraints, difficult to meet with standard numerical approaches. Therefore, parametric physics-informed neural networks (PINNs) for constitutive model calibration from full-field displacement data are investigated. In an offline stage, a parametric PINN can be trained to learn a parameterized solution of the underlying partial differential equation. In the subsequent online stage, the parametric PINN then acts as a surrogate for the parameters-to-state map in calibration. We test the proposed approach for the deterministic least-squares calibration of a linear elastic as well as a hyperelastic constitutive model from noisy synthetic displacement data. We further carry out Markov chain Monte Carlo-based Bayesian inference to quantify the uncertainty. A proper statistical evaluation of the results underlines the high accuracy of the deterministic calibration and that the estimated uncertainty is valid. Finally, we consider experimental data and show that the results are in good agreement with a finite element method-based calibration. Due to the fast evaluation of PINNs, calibration can be performed in near real-time. This advantage is particularly evident in many-query applications such as Markov chain Monte Carlo-based Bayesian inference.
- [434] arXiv:2406.02213 (replaced) [pdf, html, other]
-
Title: Random Policy Evaluation Uncovers Policies of Generative Flow NetworksSubjects: Machine Learning (cs.LG)
The Generative Flow Network (GFlowNet) is a probabilistic framework in which an agent learns a stochastic policy and flow functions to sample objects proportionally to an unnormalized reward function. A number of recent works explored connections between GFlowNets and maximum entropy (MaxEnt) RL, which modifies the standard objective of RL agents by learning an entropy-regularized objective. However, the relationship between GFlowNets and standard RL remains largely unexplored, despite the inherent similarities in their sequential decision-making nature. While GFlowNets can discover diverse solutions through specialized flow-matching objectives, connecting them can simplify their implementation through established RL principles and improve RL's diverse solution discovery capabilities. In this paper, we bridge this gap by revealing a fundamental connection between GFlowNets and one RL's most basic components -- policy evaluation. Surprisingly, we find that the value function obtained from evaluating a uniform policy is closely associated with the flow functions in GFlowNets through the lens of flow iteration under certain structural conditions. Building upon these insights, we introduce a rectified random policy evaluation (RPE) algorithm, which achieves the same reward-matching effect as GFlowNets based on simply evaluating a fixed random policy in these cases, offering a new perspective. Empirical results across extensive benchmarks demonstrate that RPE achieves competitive results compared to previous approaches, shedding light on the previously overlooked connection between (non-MaxEnt) RL and GFlowNets.
- [435] arXiv:2406.04328 (replaced) [pdf, html, other]
-
Title: The Brain's Bitter Lesson: Scaling Speech Decoding With Self-Supervised LearningComments: Published as a conference paper at ICML 2025. 16 pages, 4 figures, 4 tablesSubjects: Machine Learning (cs.LG)
The past few years have seen remarkable progress in the decoding of speech from brain activity, primarily driven by large single-subject datasets. However, due to individual variation, such as anatomy, and differences in task design and scanning hardware, leveraging data across subjects and datasets remains challenging. In turn, the field has not benefited from the growing number of open neural data repositories to exploit large-scale deep learning. To address this, we develop neuroscience-informed self-supervised objectives, together with an architecture, for learning from heterogeneous brain recordings. Scaling to nearly 400 hours of MEG data and 900 subjects, our approach shows generalisation across participants, datasets, tasks, and even to novel subjects. It achieves improvements of 15-27% over state-of-the-art models and matches surgical decoding performance with non-invasive data. These advances unlock the potential for scaling speech decoding models beyond the current frontier.
- [436] arXiv:2406.04610 (replaced) [pdf, html, other]
-
Title: Contrastive Explainable Clustering with Differential PrivacyComments: Accepted by AAMAS 2025: this https URLSubjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
This paper presents a novel approach to Explainable AI (XAI) that combines contrastive explanations with differential privacy for clustering algorithms. Focusing on k-median and k-means problems, we calculate contrastive explanations as the utility difference between original clustering and clustering with a centroid fixed to a specific data point. This method provides personalized insights into centroid placement. Our key contribution is demonstrating that these differentially private explanations achieve essentially the same utility bounds as non-private explanations. Experiments across various datasets show that our approach offers meaningful, privacy-preserving, and individually relevant explanations without significantly compromising clustering utility. This work advances privacy-aware machine learning by balancing data protection, explanation quality, and personalization in clustering tasks.
- [437] arXiv:2406.11569 (replaced) [pdf, html, other]
-
Title: Pre-Training and Personalized Fine-Tuning via Over-the-Air Federated Meta-Learning: Convergence-Generalization Trade-OffsComments: 39 pages, 8 figures, submitted for possible journal publicationSubjects: Machine Learning (cs.LG); Information Theory (cs.IT); Signal Processing (eess.SP)
For modern artificial intelligence (AI) applications such as large language models (LLMs), the training paradigm has recently shifted to pre-training followed by fine-tuning. Furthermore, owing to dwindling open repositories of data and thanks to efforts to democratize access to AI models, pre-training is expected to increasingly migrate from the current centralized deployments to federated learning (FL) implementations. Meta-learning provides a general framework in which pre-training and fine-tuning can be formalized. Meta-learning-based personalized FL (meta-pFL) moves beyond basic personalization by targeting generalization to new agents and tasks. This paper studies the generalization performance of meta-pFL for a wireless setting in which the agents participating in the pre-training phase, i.e., meta-learning, are connected via a shared wireless channel to the server. Adopting over-the-air computing, we study the trade-off between generalization to new agents and tasks, on the one hand, and convergence, on the other hand. The trade-off arises from the fact that channel impairments may enhance generalization, while degrading convergence. Extensive numerical results validate the theory.
- [438] arXiv:2407.00102 (replaced) [pdf, html, other]
-
Title: Curriculum Learning with Quality-Driven Data SelectionSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
The impressive multimodal capabilities demonstrated by OpenAI's GPT-4 have generated significant interest in the development of Multimodal Large Language Models (MLLMs). Visual instruction tuning of MLLMs with machine-generated instruction-following data has shown to enhance zero-shot capabilities across various tasks. However, there has been limited exploration into controlling the quality of the instruction this http URL methodologies for data selection in MLLMs often rely on single, unreliable scores or use downstream tasks for selection, which is time-consuming and can lead to potential overfitting on the chosen evaluation datasets. To mitigate these limitations, we propose a novel data selection methodology that utilizes image-text correlation and model perplexity to evaluate and select data of varying quality. This approach leverages the distinct distribution of these two attributes, mapping data quality into a two-dimensional space that allows for the selection of data based on their location within this distribution. By utilizing this space, we can analyze the impact of task type settings, used as prompts, on data quality. Additionally, this space can be used to construct multi-stage subsets of varying quality to facilitate curriculum learning. Our research includes comprehensive experiments conducted on various datasets. The results emphasize substantial enhancements in five commonly assessed capabilities compared to using the complete dataset. Our codes, data, and models are publicly available at: this https URL
- [439] arXiv:2407.00490 (replaced) [pdf, html, other]
-
Title: Toward Global Convergence of Gradient EM for Over-Parameterized Gaussian Mixture ModelsComments: 25 pagesSubjects: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
We study the gradient Expectation-Maximization (EM) algorithm for Gaussian Mixture Models (GMM) in the over-parameterized setting, where a general GMM with $n>1$ components learns from data that are generated by a single ground truth Gaussian distribution. While results for the special case of 2-Gaussian mixtures are well-known, a general global convergence analysis for arbitrary $n$ remains unresolved and faces several new technical barriers since the convergence becomes sub-linear and non-monotonic. To address these challenges, we construct a novel likelihood-based convergence analysis framework and rigorously prove that gradient EM converges globally with a sublinear rate $O(1/\sqrt{t})$. This is the first global convergence result for Gaussian mixtures with more than $2$ components. The sublinear convergence rate is due to the algorithmic nature of learning over-parameterized GMM with gradient EM. We also identify a new emerging technical challenge for learning general over-parameterized GMM: the existence of bad local regions that can trap gradient EM for an exponential number of steps.
- [440] arXiv:2407.01171 (replaced) [pdf, html, other]
-
Title: Neural Conditional Probability for Uncertainty QuantificationVladimir R. Kostic, Karim Lounici, Gregoire Pacreau, Pietro Novelli, Giacomo Turri, Massimiliano PontilJournal-ref: Advances in Neural Information Processing Systems (NeurIPS) 2024Subjects: Machine Learning (cs.LG); Statistics Theory (math.ST); Methodology (stat.ME); Machine Learning (stat.ML)
We introduce Neural Conditional Probability (NCP), an operator-theoretic approach to learning conditional distributions with a focus on statistical inference tasks. NCP can be used to build conditional confidence regions and extract key statistics such as conditional quantiles, mean, and covariance. It offers streamlined learning via a single unconditional training phase, allowing efficient inference without the need for retraining even when conditioning changes. By leveraging the approximation capabilities of neural networks, NCP efficiently handles a wide variety of com- plex probability distributions. We provide theoretical guarantees that ensure both optimization consistency and statistical accuracy. In experiments, we show that NCP with a 2-hidden-layer network matches or outperforms leading methods. This demonstrates that a a minimalistic architecture with a theoretically grounded loss can achieve competitive results, even in the face of more complex architectures.
- [441] arXiv:2407.06533 (replaced) [pdf, html, other]
-
Title: LETS-C: Leveraging Text Embedding for Time Series ClassificationComments: ACL 2025 (Main Conference)Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Computation and Language (cs.CL); Methodology (stat.ME)
Recent advancements in language modeling have shown promising results when applied to time series data. In particular, fine-tuning pre-trained large language models (LLMs) for time series classification tasks has achieved state-of-the-art (SOTA) performance on standard benchmarks. However, these LLM-based models have a significant drawback due to the large model size, with the number of trainable parameters in the millions. In this paper, we propose an alternative approach to leveraging the success of language modeling in the time series domain. Instead of fine-tuning LLMs, we utilize a text embedding model to embed time series and then pair the embeddings with a simple classification head composed of convolutional neural networks (CNN) and multilayer perceptron (MLP). We conducted extensive experiments on a well-established time series classification benchmark. We demonstrated LETS-C not only outperforms the current SOTA in classification accuracy but also offers a lightweight solution, using only 14.5% of the trainable parameters on average compared to the SOTA model. Our findings suggest that leveraging text embedding models to encode time series data, combined with a simple yet effective classification head, offers a promising direction for achieving high-performance time series classification while maintaining a lightweight model architecture.
- [442] arXiv:2407.08233 (replaced) [pdf, html, other]
-
Title: Hidden State Differential Private Mini-Batch Block Coordinate Descent for Multi-convexity OptimizationSubjects: Machine Learning (cs.LG)
We investigate the differential privacy (DP) guarantees under the hidden state assumption (HSA) for multi-convex problems. Recent analyses of privacy loss under the hidden state assumption have relied on strong assumptions such as convexity, thereby limiting their applicability to practical problems. In this paper, we introduce the Differential Privacy Mini-Batch Block Coordinate Descent (DP-MBCD) algorithm, accompanied by the privacy loss accounting methods under the hidden state assumption. Our proposed methods apply to a broad range of classical non-convex problems which are or can be converted to multi-convex problems, such as matrix factorization and neural network training. In addition to a tighter bound for privacy loss, our theoretical analysis is also compatible with proximal gradient descent and adaptive calibrated noise scenarios.
- [443] arXiv:2407.09577 (replaced) [pdf, html, other]
-
Title: FlashNorm: fast normalization for LLMsComments: 16 pages, 10 figuresSubjects: Machine Learning (cs.LG)
This paper presents FlashNorm, which is an exact but faster implementation of RMSNorm followed by linear layers. RMSNorm is used by many LLMs such as Llama, Mistral, and OpenELM. FlashNorm also speeds up Layer Normalization and its recently proposed replacement Dynamic Tanh (DyT) arXiv:2503.10622. FlashNorm also reduces the number of parameter tensors by simply merging the normalization weights with the weights of the next linear layer. See this https URL for code and more transformer tricks.
- [444] arXiv:2407.10454 (replaced) [pdf, html, other]
-
Title: Deflated Dynamics Value IterationSubjects: Machine Learning (cs.LG); Optimization and Control (math.OC)
The Value Iteration (VI) algorithm is an iterative procedure to compute the value function of a Markov decision process, and is the basis of many reinforcement learning (RL) algorithms as well. As the error convergence rate of VI as a function of iteration $k$ is $O(\gamma^k)$, it is slow when the discount factor $\gamma$ is close to $1$. To accelerate the computation of the value function, we propose Deflated Dynamics Value Iteration (DDVI). DDVI uses matrix splitting and matrix deflation techniques to effectively remove (deflate) the top $s$ dominant eigen-structure of the transition matrix $\mathcal{P}^{\pi}$. We prove that this leads to a $\tilde{O}(\gamma^k |\lambda_{s+1}|^k)$ convergence rate, where $\lambda_{s+1}$is $(s+1)$-th largest eigenvalue of the dynamics matrix. We then extend DDVI to the RL setting and present Deflated Dynamics Temporal Difference (DDTD) algorithm. We empirically show the effectiveness of the proposed algorithms.
- [445] arXiv:2408.03819 (replaced) [pdf, html, other]
-
Title: Leveraging Variation Theory in Counterfactual Data Augmentation for Optimized Active LearningComments: Accepted to ACL 2025 FindingsSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
Active Learning (AL) allows models to learn interactively from user feedback. This paper introduces a counterfactual data augmentation approach to AL, particularly addressing the selection of datapoints for user querying, a pivotal concern in enhancing data efficiency. Our approach is inspired by Variation Theory, a theory of human concept learning that emphasizes the essential features of a concept by focusing on what stays the same and what changes. Instead of just querying with existing datapoints, our approach synthesizes artificial datapoints that highlight potential key similarities and differences among labels using a neuro-symbolic pipeline combining large language models (LLMs) and rule-based models. Through an experiment in the example domain of text classification, we show that our approach achieves significantly higher performance when there are fewer annotated data. As the annotated training data gets larger the impact of the generated data starts to diminish showing its capability to address the cold start problem in AL. This research sheds light on integrating theories of human learning into the optimization of AL.
- [446] arXiv:2408.05787 (replaced) [pdf, html, other]
-
Title: On zero-shot learning in neural state estimation of power distribution systemsComments: 13 pages, 2 figures, associated source code available at this https URLJournal-ref: ENERGY 2025: The 15th International Conference on Smart Grids, Green Communications and IT Energy-aware Technologies. IARIA Conference (2025) 47-52Subjects: Machine Learning (cs.LG); Systems and Control (eess.SY)
This paper addresses the challenge of neural state estimation in power distribution systems. We identified a research gap in the current state of the art, which lies in the inability of models to adapt to changes in the power grid, such as loss of sensors and branch switching, in a zero-shot fashion. Based on the literature, we identified graph neural networks as the most promising class of models for this use case. Our experiments confirm their robustness to some grid changes and also show that a deeper network does not always perform better. We propose data augmentations to improve performance and conduct a comprehensive grid search of different model configurations for common zero-shot learning scenarios.
- [447] arXiv:2408.16993 (replaced) [pdf, html, other]
-
Title: A Scalable k-Medoids Clustering via Whale Optimization AlgorithmComments: 19 pages, 7 figuresSubjects: Machine Learning (cs.LG); Performance (cs.PF)
Unsupervised clustering has emerged as a critical tool for uncovering hidden patterns in vast, unlabeled datasets. However, traditional methods, such as Partitioning Around Medoids (PAM), struggle with scalability owing to their quadratic computational complexity. To address this limitation, we introduce WOA-kMedoids, a novel unsupervised clustering method that incorporates the Whale Optimization Algorithm (WOA), a nature-inspired metaheuristic inspired by the hunting strategies of humpback whales. By optimizing the centroid selection, WOA-kMedoids reduces the computational complexity from quadratic to near-linear with respect to the number of observations, enabling scalability to large datasets while maintaining high clustering accuracy. We evaluated WOA-kMedoids using 25 diverse time-series datasets from the UCR archive. Our empirical results show that WOA-kMedoids achieved a clustering performance comparable to PAM, with an average Rand Index (RI) of 0.731 compared to PAM's 0.739, outperforming PAM on 12 out of 25 datasets. While exhibiting a slightly higher runtime than PAM on small datasets (<300 observations), WOA-kMedoids outperformed PAM on larger datasets, with an average speedup of 1.7x and a maximum of 2.3x. The scalability of WOA-kMedoids, combined with its high accuracy, makes them a promising choice for unsupervised clustering in big data applications. This method has implications for efficient knowledge discovery in massive unlabeled datasets, particularly where traditional k-medoids methods are computationally infeasible, including IoT anomaly detection, biomedical signal analysis, and customer behavior clustering.
- [448] arXiv:2409.03301 (replaced) [pdf, html, other]
-
Title: ELO-Rated Sequence Rewards: Advancing Reinforcement Learning ModelsSubjects: Machine Learning (cs.LG)
Reinforcement Learning (RL) heavily relies on the careful design of the reward function. However, accurately assigning rewards to each state-action pair in Long-Term Reinforcement Learning (LTRL) tasks remains a significant challenge. As a result, RL agents are often trained under expert guidance. Inspired by the ordinal utility theory in economics, we propose a novel reward estimation algorithm: ELO-Rating based Reinforcement Learning (ERRL). This approach features two key contributions. First, it uses expert preferences over trajectories rather than cardinal rewards (utilities) to compute the ELO rating of each trajectory as its reward. Second, a new reward redistribution algorithm is introduced to alleviate training instability in the absence of a fixed anchor reward. In long-term scenarios (up to 5000 steps), where traditional RL algorithms struggle, our method outperforms several state-of-the-art baselines. Additionally, we conduct a comprehensive analysis of how expert preferences influence the results.
- [449] arXiv:2409.13280 (replaced) [pdf, html, other]
-
Title: Efficient Training of Deep Neural Operator Networks via Randomized SamplingSubjects: Machine Learning (cs.LG); Data Analysis, Statistics and Probability (physics.data-an); Machine Learning (stat.ML)
Neural operators (NOs) employ deep neural networks to learn mappings between infinite-dimensional function spaces. Deep operator network (DeepONet), a popular NO architecture, has demonstrated success in the real-time prediction of complex dynamics across various scientific and engineering applications. In this work, we introduce a random sampling technique to be adopted during the training of DeepONet, aimed at improving the generalization ability of the model, while significantly reducing the computational time. The proposed approach targets the trunk network of the DeepONet model that outputs the basis functions corresponding to the spatiotemporal locations of the bounded domain on which the physical system is defined. While constructing the loss function, DeepONet training traditionally considers a uniform grid of spatiotemporal points at which all the output functions are evaluated for each iteration. This approach leads to a larger batch size, resulting in poor generalization and increased memory demands, due to the limitations of the stochastic gradient descent (SGD) optimizer. The proposed random sampling over the inputs of the trunk net mitigates these challenges, improving generalization and reducing memory requirements during training, resulting in significant computational gains. We validate our hypothesis through three benchmark examples, demonstrating substantial reductions in training time while achieving comparable or lower overall test errors relative to the traditional training approach. Our results indicate that incorporating randomization in the trunk network inputs during training enhances the efficiency and robustness of DeepONet, offering a promising avenue for improving the framework's performance in modeling complex physical systems.
- [450] arXiv:2409.17587 (replaced) [pdf, html, other]
-
Title: Multimodal Banking Dataset: Understanding Client Needs through Event SequencesDzhambulat Mollaev, Alexander Kostin, Maria Postnova, Ivan Karpukhin, Ivan Kireev, Gleb Gusev, Andrey SavchenkoSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Financial organizations collect a huge amount of temporal (sequential) data about clients, which is typically collected from multiple sources (modalities). Despite the urgent practical need, developing deep learning techniques suitable to handle such data is limited by the absence of large open-source multi-source real-world datasets of event sequences. To fill this gap, which is mainly caused by security reasons, we present the first industrial-scale publicly available multimodal banking dataset, MBD, that contains information on more than 2M corporate clients of a large bank. Clients are represented by several data sources: 950M bank transactions, 1B geo position events, 5M embeddings of dialogues with technical support, and monthly aggregated purchases of four bank products. All entries are properly anonymized from real proprietary bank data, and the experiments confirm that our anonymization still saves all significant information for introduced downstream tasks. Moreover, we introduce a novel multimodal benchmark suggesting several important practical tasks, such as future purchase prediction and modality matching. The benchmark incorporates our MBD and two public financial datasets. We provide numerical results for the state-of-the-art event sequence modeling techniques including large language models and demonstrate the superiority of fusion baselines over single-modal techniques for each task. Thus, MBD provides a valuable resource for future research in financial applications of multimodal event sequence analysis.
HuggingFace Link: this https URL
Github Link: this https URL - [451] arXiv:2409.19078 (replaced) [pdf, other]
-
Title: Differential privacy enables fair and accurate AI-based analysis of speech disorders while protecting patient dataSoroosh Tayebi Arasteh, Mahshad Lotfinia, Paula Andrea Perez-Toro, Tomas Arias-Vergara, Mahtab Ranji, Juan Rafael Orozco-Arroyave, Maria Schuster, Andreas Maier, Seung Hee YangSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Sound (cs.SD); Audio and Speech Processing (eess.AS)
Speech pathology has impacts on communication abilities and quality of life. While deep learning-based models have shown potential in diagnosing these disorders, the use of sensitive data raises critical privacy concerns. Although differential privacy (DP) has been explored in the medical imaging domain, its application in pathological speech analysis remains largely unexplored despite the equally critical privacy concerns. To the best of our knowledge, this study is the first to investigate DP's impact on pathological speech data, focusing on the trade-offs between privacy, diagnostic accuracy, and fairness. Using a large, real-world dataset of 200 hours of recordings from 2,839 German-speaking participants, we observed a maximum accuracy reduction of 3.85% when training with DP with high privacy levels. To highlight real-world privacy risks, we demonstrated the vulnerability of non-private models to gradient inversion attacks, reconstructing identifiable speech samples and showcasing DP's effectiveness in mitigating these risks. To explore the potential generalizability across languages and disorders, we validated our approach on a dataset of Spanish-speaking Parkinson's disease patients, leveraging pretrained models from healthy English-speaking datasets, and demonstrated that careful pretraining on large-scale task-specific datasets can maintain favorable accuracy under DP constraints. A comprehensive fairness analysis revealed minimal gender bias at reasonable privacy levels but underscored the need for addressing age-related disparities. Our results establish that DP can balance privacy and utility in speech disorder detection, while highlighting unique challenges in privacy-fairness trade-offs for speech data. This provides a foundation for refining DP methodologies and improving fairness across diverse patient groups in real-world deployments.
- [452] arXiv:2410.00544 (replaced) [pdf, html, other]
-
Title: Best Practices for Multi-Fidelity Bayesian Optimization in Materials and Molecular ResearchVíctor Sabanza-Gil, Riccardo Barbano, Daniel Pacheco Gutiérrez, Jeremy S. Luterbacher, José Miguel Hernández-Lobato, Philippe Schwaller, Loïc RochSubjects: Machine Learning (cs.LG)
Multi-fidelity Bayesian Optimization (MFBO) is a promising framework to speed up materials and molecular discovery as sources of information of different accuracies are at hand at increasing cost. Despite its potential use in chemical tasks, there is a lack of systematic evaluation of the many parameters playing a role in MFBO. In this work, we provide guidelines and recommendations to decide when to use MFBO in experimental settings. We investigate MFBO methods applied to molecules and materials problems. First, we test two different families of acquisition functions in two synthetic problems and study the effect of the informativeness and cost of the approximate function. We use our implementation and guidelines to benchmark three real discovery problems and compare them against their single-fidelity counterparts. Our results may help guide future efforts to implement MFBO as a routine tool in the chemical sciences.
- [453] arXiv:2410.01104 (replaced) [pdf, html, other]
-
Title: Softmax is not Enough (for Sharp Size Generalisation)Comments: To appear at ICML 2025. 22 pages, 9 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Theory (cs.IT)
A key property of reasoning systems is the ability to make sharp decisions on their input data. For contemporary AI systems, a key carrier of sharp behaviour is the softmax function, with its capability to perform differentiable query-key lookups. It is a common belief that the predictive power of networks leveraging softmax arises from "circuits" which sharply perform certain kinds of computations consistently across many diverse inputs. However, for these circuits to be robust, they would need to generalise well to arbitrary valid inputs. In this paper, we dispel this myth: even for tasks as simple as finding the maximum key, any learned circuitry must disperse as the number of items grows at test time. We attribute this to a fundamental limitation of the softmax function to robustly approximate sharp functions with increasing problem size, prove this phenomenon theoretically, and propose adaptive temperature as an ad-hoc technique for improving the sharpness of softmax at inference time.
- [454] arXiv:2410.02628 (replaced) [pdf, html, other]
-
Title: Inverse Entropic Optimal Transport Solves Semi-supervised Learning via Data Likelihood MaximizationMikhail Persiianov, Arip Asadulaev, Nikita Andreev, Nikita Starodubcev, Dmitry Baranchuk, Anastasis Kratsios, Evgeny Burnaev, Alexander KorotinSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Learning conditional distributions $\pi^*(\cdot|x)$ is a central problem in machine learning, which is typically approached via supervised methods with paired data $(x,y) \sim \pi^*$. However, acquiring paired data samples is often challenging, especially in problems such as domain translation. This necessitates the development of $\textit{semi-supervised}$ models that utilize both limited paired data and additional unpaired i.i.d. samples $x \sim \pi^*_x$ and $y \sim \pi^*_y$ from the marginal distributions. The usage of such combined data is complex and often relies on heuristic approaches. To tackle this issue, we propose a new learning paradigm that integrates both paired and unpaired data $\textbf{seamlessly}$ using the data likelihood maximization techniques. We demonstrate that our approach also connects intriguingly with inverse entropic optimal transport (OT). This finding allows us to apply recent advances in computational OT to establish an $\textbf{end-to-end}$ learning algorithm to get $\pi^*(\cdot|x)$. In addition, we derive the universal approximation property, demonstrating that our approach can theoretically recover true conditional distributions with arbitrarily small error. Furthermore, we demonstrate through empirical tests that our method effectively learns conditional distributions using paired and unpaired data simultaneously.
- [455] arXiv:2410.02735 (replaced) [pdf, html, other]
-
Title: OOD-Chameleon: Is Algorithm Selection for OOD Generalization Learnable?Comments: ICML 2025Subjects: Machine Learning (cs.LG)
Out-of-distribution (OOD) generalization is challenging because distribution shifts come in many forms. Numerous algorithms exist to address specific settings, but choosing the right training algorithm for the right dataset without trial and error is difficult. Indeed, real-world applications often involve multiple types and combinations of shifts that are hard to analyze theoretically.
Method. This work explores the possibility of learning the selection of a training algorithm for OOD generalization. We propose a proof of concept (OOD-Chameleon) that formulates the selection as a multi-label classification over candidate algorithms, trained on a dataset of datasets representing a variety of shifts. We evaluate the ability of OOD-Chameleon to rank algorithms on unseen shifts and datasets based only on dataset characteristics, i.e., without training models first, unlike traditional model selection.
Findings. Extensive experiments show that the learned selector identifies high-performing algorithms across synthetic, vision, and language tasks. Further inspection shows that it learns non-trivial decision rules, which provide new insights into the applicability of existing algorithms. Overall, this new approach opens the possibility of better exploiting and understanding the plethora of existing algorithms for OOD generalization. - [456] arXiv:2410.03210 (replaced) [pdf, html, other]
-
Title: Tadashi: Enabling AI-Based Automated Code Generation With Guaranteed CorrectnessComments: Submitted to SC25Subjects: Machine Learning (cs.LG)
Frameworks and domain-specific languages for auto-generating code have traditionally depended on human experts to implement rigorous methods ensuring the legality of code transformations. Recently, machine learning (ML) has gained traction for generating code optimized for specific hardware targets. However, ML approaches-particularly black-box neural networks-offer no guarantees on the correctness or legality of the transformations they produce. To address this gap, we introduce Tadashi, an end-to-end system that leverages the polyhedral model to support researchers in curating datasets critical for ML-based code generation. Tadashi provides an end-to-end system capable of applying, verifying, and evaluating candidate transformations on polyhedral schedules with both reliability and practicality. We formally prove that Tadashi guarantees the legality of generated transformations, demonstrate its low runtime overhead, and showcase its broad applicability. Tadashi available at this https URL.
- [457] arXiv:2410.03960 (replaced) [pdf, html, other]
-
Title: SwiftKV: Fast Prefill-Optimized Inference with Knowledge-Preserving Model TransformationSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
LLM inference for enterprise applications, such as summarization, RAG, and code-generation, typically observe much longer prompt than generations, leading to high prefill cost and response latency. We present SwiftKV, a novel model transformation and distillation procedure targeted at reducing the prefill compute (in FLOPs) of prompt tokens while preserving high generation quality. First, SwiftKV prefills later layers' KV cache using an earlier layer's output, allowing prompt tokens to skip those later layers. Second, SwiftKV employs a lightweight knowledge-preserving distillation procedure that can adapt existing LLMs with minimal accuracy impact. Third, SwiftKV can naturally incorporate KV cache compression to improve inference performance in low-memory scenarios. Our comprehensive experiments show that SwiftKV can effectively reduce prefill computation by 25-50% across several LLM families while incurring minimum quality degradation. In the end-to-end inference serving, SwiftKV realizes up to 2x higher aggregate throughput and 60% lower time per output token. It can achieve a staggering 560 TFlops/GPU of normalized inference throughput, which translates to 16K tokens/s for Llama-3.1-70B. SwiftKV is open-sourced at this https URL.
- [458] arXiv:2410.05026 (replaced) [pdf, other]
-
Title: Active Multi-task Policy Fine-tuningSubjects: Machine Learning (cs.LG); Robotics (cs.RO)
Pre-trained generalist policies are rapidly gaining relevance in robot learning due to their promise of fast adaptation to novel, in-domain tasks. This adaptation often relies on collecting new demonstrations for a specific task of interest and applying imitation learning algorithms, such as behavioral cloning. However, as soon as several tasks need to be learned, we must decide which tasks should be demonstrated and how often? We study this multi-task problem and explore an interactive framework in which the agent adaptively selects the tasks to be demonstrated. We propose AMF (Active Multi-task Fine-tuning), an algorithm to maximize multi-task policy performance under a limited demonstration budget by collecting demonstrations yielding the largest information gain on the expert policy. We derive performance guarantees for AMF under regularity assumptions and demonstrate its empirical effectiveness to efficiently fine-tune neural policies in complex and high-dimensional environments.
- [459] arXiv:2410.05662 (replaced) [pdf, html, other]
-
Title: Federated Learning with Dynamic Client Arrival and Departure: Convergence and Rapid Adaptation via Initial Model ConstructionSubjects: Machine Learning (cs.LG)
Most federated learning (FL) approaches assume a fixed client set. However, real-world scenarios often involve clients dynamically joining or leaving the system based on their needs or interest in specific tasks. This dynamic setting introduces unique challenges: (1) the optimization objective evolves with the active client set, unlike traditional FL with a static objective; and (2) the current global model may no longer serve as an effective initialization for subsequent rounds, potentially hindering adaptation. To address these challenges, we first provide a convergence analysis under a non-convex loss with a dynamic client set, accounting for factors such as gradient noise, local training iterations, and data heterogeneity. Building on this analysis, we propose a model initialization algorithm that enables rapid adaptation to new client sets whenever clients join or leave the system. Our key idea is to compute a weighted average of previous global models, guided by gradient similarity, to prioritize models trained on data distributions that closely align with the current client set, thereby accelerating recovery from distribution shifts. This plug-and-play algorithm is designed to integrate seamlessly with existing FL methods, offering broad applicability in practice. Experimental results on diverse datasets including both image and text domains, varied label distributions, and multiple FL algorithms demonstrate the effectiveness of the proposed approach across a range of scenarios.
- [460] arXiv:2410.06820 (replaced) [pdf, html, other]
-
Title: Learning a Neural Solver for Parametric PDE to Enhance Physics-Informed MethodsLise Le Boudec, Emmanuel de Bezenac, Louis Serrano, Ramon Daniel Regueiro-Espino, Yuan Yin, Patrick GallinariSubjects: Machine Learning (cs.LG)
Physics-informed deep learning often faces optimization challenges due to the complexity of solving partial differential equations (PDEs), which involve exploring large solution spaces, require numerous iterations, and can lead to unstable training. These challenges arise particularly from the ill-conditioning of the optimization problem caused by the differential terms in the loss function. To address these issues, we propose learning a solver, i.e., solving PDEs using a physics-informed iterative algorithm trained on data. Our method learns to condition a gradient descent algorithm that automatically adapts to each PDE instance, significantly accelerating and stabilizing the optimization process and enabling faster convergence of physics-aware models. Furthermore, while traditional physics-informed methods solve for a single PDE instance, our approach extends to parametric PDEs. Specifically, we integrate the physical loss gradient with PDE parameters, allowing our method to solve over a distribution of PDE parameters, including coefficients, initial conditions, and boundary conditions. We demonstrate the effectiveness of our approach through empirical experiments on multiple datasets, comparing both training and test-time optimization performance. The code is available at this https URL.
- [461] arXiv:2410.07994 (replaced) [pdf, html, other]
-
Title: Neuroplastic Expansion in Deep Reinforcement LearningSubjects: Machine Learning (cs.LG)
The loss of plasticity in learning agents, analogous to the solidification of neural pathways in biological brains, significantly impedes learning and adaptation in reinforcement learning due to its non-stationary nature. To address this fundamental challenge, we propose a novel approach, {\it Neuroplastic Expansion} (NE), inspired by cortical expansion in cognitive science. NE maintains learnability and adaptability throughout the entire training process by dynamically growing the network from a smaller initial size to its full dimension. Our method is designed with three key components: (\textit{1}) elastic topology generation based on potential gradients, (\textit{2}) dormant neuron pruning to optimize network expressivity, and (\textit{3}) neuron consolidation via experience review to strike a balance in the plasticity-stability dilemma. Extensive experiments demonstrate that NE effectively mitigates plasticity loss and outperforms state-of-the-art methods across various tasks in MuJoCo and DeepMind Control Suite environments. NE enables more adaptive learning in complex, dynamic environments, which represents a crucial step towards transitioning deep reinforcement learning from static, one-time training paradigms to more flexible, continually adapting models.
- [462] arXiv:2410.09411 (replaced) [pdf, html, other]
-
Title: Towards the Effect of Examples on In-Context Learning: A Theoretical Case StudyComments: Accepted to Stat. Vol 14, Issue 1. Presented on JSM 2025Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
In-context learning (ICL) has emerged as a powerful capability for large language models (LLMs) to adapt to downstream tasks by leveraging a few (demonstration) examples. Despite its effectiveness, the mechanism behind ICL remains underexplored. To better understand how ICL integrates the examples with the knowledge learned by the LLM during pre-training (i.e., pre-training knowledge) and how the examples impact ICL, this paper conducts a theoretical study in binary classification tasks. In particular, we introduce a probabilistic model extending from the Gaussian mixture model to exactly quantify the impact of pre-training knowledge, label frequency, and label noise on the prediction accuracy. Based on our analysis, when the pre-training knowledge contradicts the knowledge in the examples, whether ICL prediction relies more on the pre-training knowledge or the examples depends on the number of examples. In addition, the label frequency and label noise of the examples both affect the accuracy of the ICL prediction, where the minor class has a lower accuracy, and how the label noise impacts the accuracy is determined by the specific noise level of the two classes. Extensive simulations are conducted to verify the correctness of the theoretical results, and real-data experiments also align with the theoretical insights. Our work reveals the role of pre-training knowledge and examples in ICL, offering a deeper understanding of LLMs' behaviors in classification tasks.
- [463] arXiv:2410.11674 (replaced) [pdf, html, other]
-
Title: LLM-Mixer: Multiscale Mixing in LLMs for Time Series ForecastingMd Kowsher, Md. Shohanur Islam Sobuj, Nusrat Jahan Prottasha, E. Alejandro Alanis, Ozlem Ozmen Garibay, Niloofar YousefiComments: Time series forecasting using LLMsSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Time series forecasting remains a challenging task, particularly in the context of complex multiscale temporal patterns. This study presents LLM-Mixer, a framework that improves forecasting accuracy through the combination of multiscale time-series decomposition with pre-trained LLMs (Large Language Models). LLM-Mixer captures both short-term fluctuations and long-term trends by decomposing the data into multiple temporal resolutions and processing them with a frozen LLM, guided by a textual prompt specifically designed for time-series data. Extensive experiments conducted on multivariate and univariate datasets demonstrate that LLM-Mixer achieves competitive performance, outperforming recent state-of-the-art models across various forecasting horizons. This work highlights the potential of combining multiscale analysis and LLMs for effective and scalable time-series forecasting.
- [464] arXiv:2410.12156 (replaced) [pdf, html, other]
-
Title: FragNet: A Graph Neural Network for Molecular Property Prediction with Four Levels of InterpretabilitySubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Chemical Physics (physics.chem-ph)
Molecular property prediction is essential in a variety of contemporary scientific fields, such as drug development and designing energy storage materials. Although there are many machine learning models available for this purpose, those that achieve high accuracy while also offering interpretability of predictions are uncommon. We present a graph neural network that not only matches the prediction accuracies of leading models but also provides insights on four levels of molecular substructures. This model helps identify which atoms, bonds, molecular fragments, and connections between fragments are significant for predicting a specific molecular property. Understanding the importance of connections between fragments is particularly valuable for molecules with substructures that do not connect through standard bonds. The model additionally can quantify the impact of specific fragments on the prediction, allowing the identification of fragments that may improve or degrade a property value. These interpretable features are essential for deriving scientific insights from the model's learned relationships between molecular structures and properties.
- [465] arXiv:2410.13097 (replaced) [pdf, html, other]
-
Title: Communication-Efficient and Tensorized Federated Fine-Tuning of Large Language ModelsSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Parameter-efficient fine-tuning (PEFT) methods typically assume that Large Language Models (LLMs) are trained on data from a single device or client. However, real-world scenarios often require fine-tuning these models on private data distributed across multiple devices. Federated Learning (FL) offers an appealing solution by preserving user privacy, as sensitive data remains on local devices during training. Nonetheless, integrating PEFT methods into FL introduces two main challenges: communication overhead and data heterogeneity. In this paper, we introduce FedTT and FedTT+, methods for adapting LLMs by integrating tensorized adapters into client-side models' encoder/decoder blocks. FedTT is versatile and can be applied to both cross-silo FL and large-scale cross-device FL. FedTT+, an extension of FedTT tailored for cross-silo FL, enhances robustness against data heterogeneity by adaptively freezing portions of tensor factors, further reducing the number of trainable parameters. Experiments on BERT and LLaMA models demonstrate that our proposed methods successfully address data heterogeneity challenges and perform on par or even better than existing federated PEFT approaches while achieving up to 10$\times$ reduction in communication cost.
- [466] arXiv:2410.13248 (replaced) [pdf, html, other]
-
Title: Disentangling Likes and Dislikes in Personalized Generative Explainable RecommendationRyotaro Shimizu, Takashi Wada, Yu Wang, Johannes Kruse, Sean O'Brien, Sai HtaungKham, Linxin Song, Yuya Yoshikawa, Yuki Saito, Fugee Tsung, Masayuki Goto, Julian McAuleyComments: This manuscript has been accepted for presentation at The Web Conference (WWW) 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR)
Recent research on explainable recommendation generally frames the task as a standard text generation problem, and evaluates models simply based on the textual similarity between the predicted and ground-truth explanations. However, this approach fails to consider one crucial aspect of the systems: whether their outputs accurately reflect the users' (post-purchase) sentiments, i.e., whether and why they would like and/or dislike the recommended items. To shed light on this issue, we introduce new datasets and evaluation methods that focus on the users' sentiments. Specifically, we construct the datasets by explicitly extracting users' positive and negative opinions from their post-purchase reviews using an LLM, and propose to evaluate systems based on whether the generated explanations 1) align well with the users' sentiments, and 2) accurately identify both positive and negative opinions of users on the target items. We benchmark several recent models on our datasets and demonstrate that achieving strong performance on existing metrics does not ensure that the generated explanations align well with the users' sentiments. Lastly, we find that existing models can provide more sentiment-aware explanations when the users' (predicted) ratings for the target items are directly fed into the models as input. The datasets and benchmark implementation are available at: this https URL.
- [467] arXiv:2410.17933 (replaced) [pdf, html, other]
-
Title: Multi-Continental Healthcare Modelling Using Blockchain-Enabled Federated LearningComments: Camera Ready Version, Accepted by IEEE Global Blockchain Conference, 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
One of the biggest challenges of building artificial intelligence (AI) model in healthcare area is the data sharing. Since healthcare data is private, sensitive, and heterogeneous, collecting sufficient data for modelling is exhausted, costly, and sometimes impossible. In this paper, we propose a framework for global healthcare modelling using datasets from multi-continents (Europe, North America and Asia) while without sharing the local datasets, and choose glucose management as a study model to verify its effectiveness. Technically, blockchain-enabled federated learning is implemented with adaption to make it meet with the privacy and safety requirements of healthcare data, meanwhile rewards honest participation and penalize malicious activities using its on-chain incentive mechanism. Experimental results show that the proposed framework is effective, efficient, and privacy preserved. Its prediction accuracy is much better than the models trained from limited personal data and is similar to, and even slightly better than, the results from a centralized dataset. This work paves the way for international collaborations on healthcare projects, where additional data is crucial for reducing bias and providing benefits to humanity.
- [468] arXiv:2410.22954 (replaced) [pdf, html, other]
-
Title: Retrieval-Augmented Generation with Estimation of Source ReliabilitySubjects: Machine Learning (cs.LG)
Retrieval-augmented generation (RAG) addresses key limitations of large language models (LLMs), such as hallucinations and outdated knowledge, by incorporating external databases. These databases typically consult multiple sources to encompass up-to-date and various information. However, standard RAG methods often overlook the heterogeneous source reliability in the multi-source database and retrieve documents solely based on relevance, making them prone to propagating misinformation. To address this, we propose Reliability-Aware RAG (RA-RAG) which estimates the reliability of multiple sources and incorporates this information into both retrieval and aggregation processes. Specifically, it iteratively estimates source reliability and true answers for a set of queries with no labelling. Then, it selectively retrieves relevant documents from a few of reliable sources and aggregates them using weighted majority voting, where the selective retrieval ensures scalability while not compromising the performance. We also introduce a benchmark designed to reflect real-world scenarios with heterogeneous source reliability and demonstrate the effectiveness of RA-RAG compared to a set of baselines.
- [469] arXiv:2411.00524 (replaced) [pdf, html, other]
-
Title: Comparison-based Active Preference Learning for Multi-dimensional PersonalizationSubjects: Machine Learning (cs.LG)
Large language models (LLMs) have shown remarkable success, but aligning them with human preferences remains a core challenge. As individuals have their own, multi-dimensional preferences, recent studies have explored multi-dimensional personalization, which aims to enable models to generate responses personalized to explicit preferences. However, human preferences are often implicit and thus difficult to articulate, limiting the direct application of this approach. To bridge this gap, we propose Active Multi-dimensional Preference Learning (AMPLe), designed to capture implicit user preferences from interactively collected comparative feedback. Building on Bayesian inference, our work introduces a modified posterior update procedure to mitigate estimation bias and potential noise in comparisons. Also, inspired by generalized binary search, we employ an active query selection strategy to minimize the number of required comparisons by a user. Through theoretical analysis and experiments on language generation tasks, we demonstrate feedback efficiency and effectiveness of our framework in personalizing model responses. Our code is publicly available at this https URL .
- [470] arXiv:2411.01341 (replaced) [pdf, other]
-
Title: Convolutional Filtering with RKHS AlgebrasSubjects: Machine Learning (cs.LG); Signal Processing (eess.SP)
In this paper, we develop a generalized theory of convolutional signal processing and neural networks for Reproducing Kernel Hilbert Spaces (RKHS). Leveraging the theory of algebraic signal processing (ASP), we show that any RKHS allows the formal definition of multiple algebraic convolutional models. We show that any RKHS induces algebras whose elements determine convolutional operators acting on RKHS elements. This approach allows us to achieve scalable filtering and learning as a byproduct of the convolutional model, and simultaneously take advantage of the well-known benefits of processing information in an RKHS. To emphasize the generality and usefulness of our approach, we show how algebraic RKHS can be used to define convolutional signal models on groups, graphons, and traditional Euclidean signal spaces. Furthermore, using algebraic RKHS models, we build convolutional networks, formally defining the notion of pointwise nonlinearities and deriving explicit expressions for the training. Such derivations are obtained in terms of the algebraic representation of the RKHS. We present a set of numerical experiments on real data in which wireless coverage is predicted from measurements captured by unmaned aerial vehicles. This particular real-life scenario emphasizes the benefits of the convolutional RKHS models in neural networks compared to fully connected and standard convolutional operators.
- [471] arXiv:2411.04760 (replaced) [pdf, html, other]
-
Title: Zero-Shot Temporal Resolution Domain Adaptation for Spiking Neural NetworksSubjects: Machine Learning (cs.LG)
Spiking Neural Networks (SNNs) are biologically-inspired deep neural networks that efficiently extract temporal information while offering promising gains in terms of energy efficiency and latency when deployed on neuromorphic devices. However, SNN model parameters are sensitive to temporal resolution, leading to significant performance drops when the temporal resolution of target data at the edge is not the same with that of the pre-deployment source data used for training, especially when fine-tuning is not possible at the edge. To address this challenge, we propose three novel domain adaptation methods for adapting neuron parameters to account for the change in time resolution without re-training on target time-resolution. The proposed methods are based on a mapping between neuron dynamics in SNNs and State Space Models (SSMs); and are applicable to general neuron models. We evaluate the proposed methods under spatio-temporal data tasks, namely the audio keyword spotting datasets SHD and MSWC as well as the image classification NMINST dataset. Our methods provide an alternative to - and in majority of the cases significantly outperform - the existing reference method that simply scales the time constant. Moreover, our results show that high accuracy on high temporal resolution data can be obtained by time efficient training on lower temporal resolution data and model adaptation.
- [472] arXiv:2411.09749 (replaced) [pdf, html, other]
-
Title: RenderBender: A Survey on Adversarial Attacks Using Differentiable RenderingMatthew Hull, Haoran Wang, Matthew Lau, Alec Helbling, Mansi Phute, Chao Zhang, Zsolt Kira, Willian Lunardi, Martin Andreoni, Wenke Lee, Polo ChauComments: 9 pages, 1 figure, 2 tables, IJCAI '25 Survey TrackSubjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV)
Differentiable rendering techniques like Gaussian Splatting and Neural Radiance Fields have become powerful tools for generating high-fidelity models of 3D objects and scenes. Their ability to produce both physically plausible and differentiable models of scenes are key ingredient needed to produce physically plausible adversarial attacks on DNNs. However, the adversarial machine learning community has yet to fully explore these capabilities, partly due to differing attack goals (e.g., misclassification, misdetection) and a wide range of possible scene manipulations used to achieve them (e.g., alter texture, mesh). This survey contributes the first framework that unifies diverse goals and tasks, facilitating easy comparison of existing work, identifying research gaps, and highlighting future directions - ranging from expanding attack goals and tasks to account for new modalities, state-of-the-art models, tools, and pipelines, to underscoring the importance of studying real-world threats in complex scenes.
- [473] arXiv:2411.09837 (replaced) [pdf, html, other]
-
Title: Real-time Adapting Routing (RAR): Improving Efficiency Through Continuous Learning in Software Powered by Layered Foundation ModelsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
To balance the quality and inference cost of a Foundation Model (FM, such as large language models (LLMs)) powered software, people often opt to train a routing model that routes requests to FMs with different sizes and capabilities. Existing routing models rely on learning the optimal routing decision from carefully curated data, require complex computations to be updated, and do not consider the potential evolution of weaker FMs. In this paper, we propose Real-time Adaptive Routing (RAR), an approach to continuously adapt FM routing decisions while using guided in-context learning to enhance the capabilities of weaker FM. The goal is to reduce reliance on stronger, more expensive FMs. We evaluate our approach on different subsets of the popular MMLU benchmark. Over time, our approach routes 50.2% fewer requests to computationally expensive models while maintaining around 90.5% of the general response quality. In addition, the guides generated from stronger models have shown intra-domain generalization and led to a better quality of responses compared to an equivalent approach with a standalone weaker FM.
- [474] arXiv:2411.10958 (replaced) [pdf, html, other]
-
Title: SageAttention2: Efficient Attention with Thorough Outlier Smoothing and Per-thread INT4 QuantizationComments: @inproceedings{zhang2024sageattention2, title={Sageattention2: Efficient attention with thorough outlier smoothing and per-thread int4 quantization}, author={Zhang, Jintao and Huang, Haofeng and Zhang, Pengle and Wei, Jia and Zhu, Jun and Chen, Jianfei}, booktitle={International Conference on Machine Learning (ICML)}, year={2025} }Journal-ref: Proceedings of the 42nd International Conference on Machine Learning, PMLR 267, 2025 (ICML 2025)Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Neural and Evolutionary Computing (cs.NE); Performance (cs.PF)
Although quantization for linear layers has been widely used, its application to accelerate the attention process remains limited. To further enhance the efficiency of attention computation compared to SageAttention while maintaining precision, we propose SageAttention2, which utilizes significantly faster 4-bit matrix multiplication (Matmul) alongside additional precision-enhancing techniques. First, we propose to quantize matrices $(Q, K)$ to INT4 in a hardware-friendly thread-level granularity and quantize matrices $(\widetilde P, V)$ to FP8. Second, we propose a method to smooth $Q$, enhancing the accuracy of INT4 $QK^\top$. Third, we propose a two-level accumulation strategy for $\widetilde PV$ to enhance the accuracy of FP8 $\widetilde PV$. The operations per second (OPS) of SageAttention2 surpass FlashAttention2 and xformers by about 3x and 4.5x on RTX4090, respectively. Moreover, SageAttention2 matches the speed of FlashAttention3(fp8) on the Hopper GPUs, while delivering much higher accuracy. Comprehensive experiments confirm that our approach incurs negligible end-to-end metrics loss across diverse models, including those for language, image, and video generation. The code is available at this https URL.
- [475] arXiv:2411.13899 (replaced) [pdf, html, other]
-
Title: Schemato - An LLM for Netlist-to-Schematic ConversionRyoga Matsuo, Stefan Uhlich, Arun Venkitaraman, Andrea Bonetti, Chia-Yu Hsieh, Ali Momeni, Lukas Mauch, Augusto Capone, Eisaku Ohbuchi, Lorenzo ServadeiSubjects: Machine Learning (cs.LG); Hardware Architecture (cs.AR)
Machine learning models are advancing circuit design, particularly in analog circuits. They typically generate netlists that lack human interpretability. This is a problem as human designers heavily rely on the interpretability of circuit diagrams or schematics to intuitively understand, troubleshoot, and develop designs. Hence, to integrate domain knowledge effectively, it is crucial to translate ML-generated netlists into interpretable schematics quickly and accurately. We propose Schemato, a large language model (LLM) for netlist-to-schematic conversion. In particular, we consider our approach in converting netlists to .asc files, text-based schematic description used in LTSpice. Experiments on our circuit dataset show that Schemato achieves up to 76% compilation success rate, surpassing 63% scored by the state-of-the-art LLMs. Furthermore, our experiments show that Schemato generates schematics with an average graph edit distance score and mean structural similarity index measure, scaled by the compilation success rate that are 1.8x and 4.3x higher than the best performing LLMs respectively, demonstrating its ability to generate schematics that are more accurately connected and are closer to the reference human design.
- [476] arXiv:2411.15645 (replaced) [pdf, html, other]
-
Title: MC-NEST: Enhancing Mathematical Reasoning in Large Language Models leveraging a Monte Carlo Self-Refine TreeSubjects: Machine Learning (cs.LG)
Mathematical reasoning presents significant challenges for large language models (LLMs). To enhance their capabilities, we propose Monte Carlo Self-Refine Tree (MC-NEST), an extension of Monte Carlo Tree Search that integrates LLM-based self-refinement and self-evaluation for improved decision-making in complex reasoning tasks. MC-NEST balances exploration and exploitation using Upper Confidence Bound (UCT) scores combined with diverse selection policies. Through iterative critique and refinement, LLMs learn to reason more strategically. Empirical results demonstrate that MC-NEST with an importance sampling policy substantially improves GPT-4o's performance, achieving state-of-the-art pass@1 scores on Olympiad-level benchmarks. Specifically, MC-NEST attains a pass@1 of 38.6 on AIME and 12.6 on MathOdyssey. The solution quality for MC-NEST using GPT-4o and Phi-3-mini reaches 84.0\% and 82.08\%, respectively, indicating robust consistency across different LLMs. MC-NEST performs strongly across Algebra, Geometry, and Number Theory, benefiting from its ability to handle abstraction, logical deduction, and multi-step reasoning -- core skills in mathematical problem solving.
- [477] arXiv:2411.17332 (replaced) [pdf, html, other]
-
Title: On the Generalization of Handwritten Text Recognition ModelsSubjects: Machine Learning (cs.LG)
Recent advances in Handwritten Text Recognition (HTR) have led to significant reductions in transcription errors on standard benchmarks under the i.i.d. assumption, thus focusing on minimizing in-distribution (ID) errors. However, this assumption does not hold in real-world applications, which has motivated HTR research to explore Transfer Learning and Domain Adaptation techniques. In this work, we investigate the unaddressed limitations of HTR models in generalizing to out-of-distribution (OOD) data. We adopt the challenging setting of Domain Generalization, where models are expected to generalize to OOD data without any prior access. To this end, we analyze 336 OOD cases from eight state-of-the-art HTR models across seven widely used datasets, spanning five languages. Additionally, we study how HTR models leverage synthetic data to generalize. We reveal that the most significant factor for generalization lies in the textual divergence between domains, followed by visual divergence. We demonstrate that the error of HTR models in OOD scenarios can be reliably estimated, with discrepancies falling below 10 points in 70\% of cases. We identify the underlying limitations of HTR models, laying the foundation for future research to address this challenge.
- [478] arXiv:2412.08174 (replaced) [pdf, html, other]
-
Title: Can Graph Neural Networks Learn Language with Extremely Weak Text Supervision?Comments: ACL 2025 Main Conference, 27 pagesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Social and Information Networks (cs.SI)
While great success has been achieved in building vision models with Contrastive Language-Image Pre-training (CLIP) over internet-scale image-text pairs, building transferable Graph Neural Networks (GNNs) with CLIP pipeline is challenging because of the scarcity of labeled data and text supervision, different levels of downstream tasks, and the conceptual gaps between domains. In this work, to address these issues, we propose a multi-modal prompt learning paradigm to effectively adapt pre-trained GNN to downstream tasks and data, given only a few semantically labeled samples, each with extremely weak text supervision. Our new paradigm embeds the graphs directly in the same space as the Large Language Models (LLMs) by learning both graph prompts and text prompts simultaneously. We demonstrate the superior performance of our paradigm in few-shot, multi-task-level, and cross-domain settings. Moreover, we build the first CLIP-style zero-shot classification prototype that can generalize GNNs to unseen classes with extremely weak text supervision. The code is available at this https URL.
- [479] arXiv:2412.08559 (replaced) [pdf, html, other]
-
Title: Underestimated Privacy Risks for Minority Populations in Large Language Model UnlearningRongzhe Wei, Mufei Li, Mohsen Ghassemi, Eleonora Kreačić, Yifan Li, Xiang Yue, Bo Li, Vamsi K. Potluru, Pan Li, Eli ChienSubjects: Machine Learning (cs.LG)
Large Language Models (LLMs) embed sensitive, human-generated data, prompting the need for unlearning methods. Although certified unlearning offers strong privacy guarantees, its restrictive assumptions make it unsuitable for LLMs, giving rise to various heuristic approaches typically assessed through empirical evaluations. These standard evaluations randomly select data for removal, apply unlearning techniques, and use membership inference attacks (MIAs) to compare unlearned models against models retrained without the removed data. However, to ensure robust privacy protections for every data point, it is essential to account for scenarios in which certain data subsets face elevated risks. Prior research suggests that outliers, particularly including data tied to minority groups, often exhibit higher memorization propensity which indicates they may be more difficult to unlearn. Building on these insights, we introduce a complementary, minority-aware evaluation framework to highlight blind spots in existing frameworks. We substantiate our findings with carefully designed experiments, using canaries with personally identifiable information (PII) to represent these minority subsets and demonstrate that they suffer at least 20% higher privacy leakage across various unlearning methods, MIAs, datasets, and LLM scales. Our proposed minority-aware evaluation framework marks an essential step toward more equitable and comprehensive assessments of LLM unlearning efficacy.
- [480] arXiv:2412.09059 (replaced) [pdf, html, other]
-
Title: Go With the Flow: Fast Diffusion for Gaussian Mixture ModelsSubjects: Machine Learning (cs.LG)
Schrodinger Bridges (SBs) are diffusion processes that steer, in finite time, a given initial distribution to another final one while minimizing a suitable cost functional. Although various methods for computing SBs have recently been proposed in the literature, most of these approaches require computationally expensive training schemes, even for solving low-dimensional problems. In this work, we propose an analytic parametrization of a set of feasible policies for steering the distribution of a dynamical system from one Gaussian Mixture Model (GMM) to another. Instead of relying on standard non-convex optimization techniques, the optimal policy within the set can be approximated as the solution of a low-dimensional linear program whose dimension scales linearly with the number of components in each mixture. The proposed method generalizes naturally to more general classes of dynamical systems, such as controllable linear time-varying systems, enabling efficient solutions to multi-marginal momentum SB between GMMs, a challenging distribution interpolation problem. We showcase the potential of this approach in low-to-moderate dimensional problems such as image-to-image translation in the latent space of an autoencoder, learning of cellular dynamics using multi-marginal momentum SB problems, and various other examples. We also test our approach on an Entropic Optimal Transport (EOT) benchmark problem and show that it outperforms state-of-the-art methods in cases where the boundary distributions are mixture models while requiring virtually no training.
- [481] arXiv:2412.10208 (replaced) [pdf, html, other]
-
Title: Efficient Generative Modeling with Residual Vector Quantization-Based TokensComments: ICML 2025Subjects: Machine Learning (cs.LG)
We introduce ResGen, an efficient Residual Vector Quantization (RVQ)-based generative model for high-fidelity generation with fast sampling. RVQ improves data fidelity by increasing the number of quantization steps, referred to as depth, but deeper quantization typically increases inference steps in generative models. To address this, ResGen directly predicts the vector embedding of collective tokens rather than individual ones, ensuring that inference steps remain independent of RVQ depth. Additionally, we formulate token masking and multi-token prediction within a probabilistic framework using discrete diffusion and variational inference. We validate the efficacy and generalizability of the proposed method on two challenging tasks across different modalities: conditional image generation on ImageNet 256x256 and zero-shot text-to-speech synthesis. Experimental results demonstrate that ResGen outperforms autoregressive counterparts in both tasks, delivering superior performance without compromising sampling speed. Furthermore, as we scale the depth of RVQ, our generative models exhibit enhanced generation fidelity or faster sampling speeds compared to similarly sized baseline models.
- [482] arXiv:2412.13667 (replaced) [pdf, html, other]
-
Title: Exploring Multi-Modal Data with Tool-Augmented LLM Agents for Precise Causal DiscoverySubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Methodology (stat.ME)
Causal discovery is an imperative foundation for decision-making across domains, such as smart health, AI for drug discovery and AIOps. Traditional statistical causal discovery methods, while well-established, predominantly rely on observational data and often overlook the semantic cues inherent in cause-and-effect relationships. The advent of Large Language Models (LLMs) has ushered in an affordable way of leveraging the semantic cues for knowledge-driven causal discovery, but the development of LLMs for causal discovery lags behind other areas, particularly in the exploration of multi-modal data. To bridge the gap, we introduce MATMCD, a multi-agent system powered by tool-augmented LLMs. MATMCD has two key agents: a Data Augmentation agent that retrieves and processes modality-augmented data, and a Causal Constraint agent that integrates multi-modal data for knowledge-driven reasoning. The proposed design of the inner-workings ensures successful cooperation of the agents. Our empirical study across seven datasets suggests the significant potential of multi-modality enhanced causal discovery.
- [483] arXiv:2412.14297 (replaced) [pdf, html, other]
-
Title: Distributionally Robust Policy Learning under Concept DriftsComments: Poster at ICML2025Journal-ref: Proceedings of the 42nd International Conference on Machine Learning, Vancouver, Canada. PMLR 267, 2025Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Distributionally robust policy learning aims to find a policy that performs well under the worst-case distributional shift, and yet most existing methods for robust policy learning consider the worst-case joint distribution of the covariate and the outcome. The joint-modeling strategy can be unnecessarily conservative when we have more information on the source of distributional shifts. This paper studies a more nuanced problem -- robust policy learning under the concept drift, when only the conditional relationship between the outcome and the covariate changes. To this end, we first provide a doubly-robust estimator for evaluating the worst-case average reward of a given policy under a set of perturbed conditional distributions. We show that the policy value estimator enjoys asymptotic normality even if the nuisance parameters are estimated with a slower-than-root-$n$ rate. We then propose a learning algorithm that outputs the policy maximizing the estimated policy value within a given policy class $\Pi$, and show that the sub-optimality gap of the proposed algorithm is of the order $\kappa(\Pi)n^{-1/2}$, where $\kappa(\Pi)$ is the entropy integral of $\Pi$ under the Hamming distance and $n$ is the sample size. A matching lower bound is provided to show the optimality of the rate. The proposed methods are implemented and evaluated in numerical studies, demonstrating substantial improvement compared with existing benchmarks.
- [484] arXiv:2412.16318 (replaced) [pdf, html, other]
-
Title: Principal-Agent Bandit Games with Self-Interested and Exploratory Learning AgentsComments: 48 pages, ICML 2025Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
We study the repeated principal-agent bandit game, where the principal indirectly interacts with the unknown environment by proposing incentives for the agent to play arms. Most existing work assumes the agent has full knowledge of the reward means and always behaves greedily, but in many online marketplaces, the agent needs to learn the unknown environment and sometimes explore. Motivated by such settings, we model a self-interested learning agent with exploration behaviors who iteratively updates reward estimates and either selects an arm that maximizes the estimated reward plus incentive or explores arbitrarily with a certain probability. As a warm-up, we first consider a self-interested learning agent without exploration. We propose algorithms for both i.i.d. and linear reward settings with bandit feedback in a finite horizon $T$, achieving regret bounds of $\widetilde{O}(\sqrt{T})$ and $\widetilde{O}( T^{2/3} )$, respectively. Specifically, these algorithms are established upon a novel elimination framework coupled with newly-developed search algorithms which accommodate the uncertainty arising from the learning behavior of the agent. We then extend the framework to handle the exploratory learning agent and develop an algorithm to achieve a $\widetilde{O}(T^{2/3})$ regret bound in i.i.d. reward setup by enhancing the robustness of our elimination framework to the potential agent exploration. Finally, when reducing our agent behaviors to the one studied in (Dogan et al., 2023a), we propose an algorithm based on our robust framework, which achieves a $\widetilde{O}(\sqrt{T})$ regret bound, significantly improving upon their $\widetilde{O}(T^{11/12})$ bound.
- [485] arXiv:2412.18730 (replaced) [pdf, html, other]
-
Title: Elucidating Flow Matching ODE Dynamics with Respect to Data GeometriesComments: Accepted to ICML 2025Subjects: Machine Learning (cs.LG)
Flow matching (FM) models extend ODE sampler based diffusion models into a general framework, significantly reducing sampling steps through learned vector fields. However, the theoretical understanding of FM models, particularly how their sample trajectories interact with underlying data geometry, remains underexplored. A rigorous theoretical analysis of FM ODE is essential for sample quality, stability, and broader applicability. In this paper, we advance the theory of FM models through a comprehensive analysis of sample trajectories. Central to our theory is the discovery that the denoiser, a key component of FM models, guides ODE dynamics through attracting and absorbing behaviors that adapt to the data geometry. We identify and analyze the three stages of ODE evolution: in the initial and intermediate stages, trajectories move toward the mean and local clusters of the data. At the terminal stage, we rigorously establish the convergence of FM ODE under weak assumptions, addressing scenarios where the data lie on a low-dimensional submanifold-cases that previous results could not handle. Our terminal stage analysis offers insights into the memorization phenomenon and establishes equivariance properties of FM ODEs. These findings bridge critical gaps in understanding flow matching models, with practical implications for optimizing sampling strategies and architectures guided by the intrinsic geometry of data.
- [486] arXiv:2501.00659 (replaced) [pdf, html, other]
-
Title: Why Are Positional Encodings Nonessential for Deep Autoregressive Transformers? Revisiting a PetroglyphComments: Accepted to ACL 2025 Findings, Short paperSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Do autoregressive Transformer language models require explicit positional encodings (PEs)? The answer is 'no' provided they have more than one layer -- they can distinguish sequences with permuted tokens without the need for explicit PEs. This follows from the fact that a cascade of (permutation invariant) set processors can collectively exhibit sequence-sensitive behavior in the autoregressive setting. This property has been known since early efforts (contemporary with GPT-2) adopting the Transformer for language modeling. However, this result does not appear to have been well disseminated, leading to recent rediscoveries. This may be partially due to a sudden growth of the language modeling community after the advent of GPT-2/3, but perhaps also due to the lack of a clear explanation in prior work, despite being commonly understood by practitioners in the past. Here we review the long-forgotten explanation why explicit PEs are nonessential for multi-layer autoregressive Transformers (in contrast, one-layer models require PEs to discern order information of their inputs), as well as the origin of this result, and hope to re-establish it as a common knowledge.
- [487] arXiv:2501.02379 (replaced) [pdf, html, other]
-
Title: TensorGRaD: Tensor Gradient Robust Decomposition for Memory-Efficient Neural Operator TrainingSebastian Loeschcke, David Pitt, Robert Joseph George, Jiawei Zhao, Cheng Luo, Yuandong Tian, Jean Kossaifi, Anima AnandkumarSubjects: Machine Learning (cs.LG)
Scientific problems require resolving multi-scale phenomena across different resolutions and learning solution operators in infinite-dimensional function spaces. Neural operators provide a powerful framework for this, using tensor-parameterized layers to capture complex, multi-dimensional relationships. However, scaling neural operators to high-resolution problems leads to significant computational demands, making the training of industrial-scale models prohibitive. In this work, we introduce \textbf{TensorGRaD}, a novel method that directly addresses the memory challenges associated with optimizing large tensor-structured weights. Our approach, based on a \texit{robust tensor decomposition}, factorizes gradients as the sum of a low-rank tensor and a sparse one to efficiently capture information within optimizer states, including outliers. Additionally, we provide a recipe for mixed precision training of TensorGRaD, achieving further memory savings without sacrificing accuracy. We showcase the effectiveness of TensorGRaD on Fourier Neural Operators, a class of models crucial for solving partial differential equations (PDE). We provide theoretical guarantees for TensorGRaD, demonstrating its fundamental advantage over matrix-based gradient compression methods. We empirically demonstrate large improvements across various PDE tasks, including the challenging turbulent Navier-Stokes case at a Reynolds number of $10^5$. TensorGRaD reduces total memory usage by over $50\%$ while maintaining and sometimes even improving accuracy.
- [488] arXiv:2501.05000 (replaced) [pdf, html, other]
-
Title: Load Forecasting for Households and Energy Communities: Are Deep Learning Models Worth the Effort?Lukas Moosbrugger, Valentin Seiler, Philipp Wohlgenannt, Sebastian Hegenbart, Sashko Ristov, Elias Eder, Peter KepplingerSubjects: Machine Learning (cs.LG)
Energy communities (ECs) play a key role in enabling local demand shifting and enhancing self-sufficiency, as energy systems transition toward decentralized structures with high shares of renewable generation. To optimally operate them, accurate short-term load forecasting is essential, particularly for implementing demand-side management strategies. With the recent rise of deep learning methods, data-driven forecasting has gained significant attention, however, it remains insufficiently explored in many practical contexts. Therefore, this study evaluates the effectiveness of state-of-the-art deep learning models -- including LSTM, xLSTM, and Transformer architectures -- compared to traditional benchmarks such as K-Nearest Neighbors (KNN) and persistence forecasting, across varying community size, historical data availability, and model complexity. Additionally, we assess the benefits of transfer learning using publicly available synthetic load profiles. On average, transfer learning improves the normalized mean absolute error by 1.97%pt when only two months of training data are available. Interestingly, for less than six months of training data, simple persistence models outperform deep learning architectures in forecast accuracy. The practical value of improved forecasting is demonstrated using a mixed-integer linear programming optimization for ECs with a shared battery energy storage system. The most accurate deep learning model achieves an average reduction in financial energy costs of 8.06%. Notably, a simple KNN approach achieves average savings of 8.01%, making it a competitive and robust alternative. All implementations are publicly available to facilitate reproducibility. These findings offer actionable insights for ECs, and they highlight when the additional complexity of deep learning is warranted by performance gains.
- [489] arXiv:2501.08219 (replaced) [pdf, html, other]
-
Title: Investigating Energy Efficiency and Performance Trade-offs in LLM Inference Across Tasks and DVFS SettingsSubjects: Machine Learning (cs.LG)
Large Language Models (LLMs) have demonstrated remarkable performance across a wide range of natural language processing (NLP) tasks, leading to widespread adoption in both research and industry. However, their inference workloads are computationally and energy intensive, raising concerns about sustainability and environmental impact. As LLMs continue to scale, it becomes essential to identify and optimize the factors that influence their runtime efficiency without compromising performance. In this work, we systematically investigate the energy-performance trade-offs of LLMs during inference. We benchmark models of varying sizes and architectures, including Falcon-7B, Mistral-7B-v0.1, LLaMA-3.2-1B, LLaMA-3.2-3B, and GPT-Neo-2.7B, across tasks such as question answering, commonsense reasoning, and factual generation. We analyze the effect of input characteristics, such as sequence length, entropy, named entity density and so on. Furthermore, we examine the impact of hardware-level optimizations through Dynamic Voltage and Frequency Scaling (DVFS), measuring how different GPU clock settings affect latency and power consumption. Our empirical findings show that model architecture, input complexity, and clock configuration significantly influence inference efficiency. By correlating input features with energy metrics and evaluating DVFS behavior, we identify practical strategies that reduce energy consumption by up to 30% while preserving model quality. This study provides actionable insights for designing energy-efficient and sustainable LLM inference systems.
- [490] arXiv:2501.14809 (replaced) [pdf, html, other]
-
Title: Evaluation of Seismic Artificial Intelligence with UncertaintySubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Geophysics (physics.geo-ph)
Artificial intelligence has transformed the seismic community with deep learning models (DLMs) that are trained to complete specific tasks within workflows. However, there is still lack of robust evaluation frameworks for evaluating and comparing DLMs. We address this gap by designing an evaluation framework that jointly incorporates two crucial aspects: performance uncertainty and learning efficiency. To target these aspects, we meticulously construct the training, validation, and test splits using a clustering method tailored to seismic data and enact an expansive training design to segregate performance uncertainty arising from stochastic training processes and random data sampling. The framework's ability to guard against misleading declarations of model superiority is demonstrated through evaluation of PhaseNet [1], a popular seismic phase picking DLM, under 3 training approaches. Our framework helps practitioners choose the best model for their problem and set performance expectations by explicitly analyzing model performance with uncertainty at varying budgets of training data.
- [491] arXiv:2501.15278 (replaced) [pdf, html, other]
-
Title: PIP: Perturbation-based Iterative Pruning for Large Language ModelsSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
The rapid increase in the parameter counts of Large Language Models (LLMs), reaching billions or even trillions, presents significant challenges for their practical deployment, particularly in resource-constrained environments. To ease this issue, we propose PIP (Perturbation-based Iterative Pruning), a novel double-view structured pruning method to optimize LLMs, which combines information from two different views: the unperturbed view and the perturbed view. With the calculation of gradient differences, PIP iteratively prunes those that struggle to distinguish between these two views. Our experiments show that PIP reduces the parameter count by approximately 20% while retaining over 85% of the original model's accuracy across varied benchmarks. In some cases, the performance of the pruned model is within 5% of the unpruned version, demonstrating PIP's ability to preserve key aspects of model effectiveness. Moreover, PIP consistently outperforms existing state-of-the-art (SOTA) structured pruning methods, establishing it as a leading technique for optimizing LLMs in environments with constrained resources.
- [492] arXiv:2501.15361 (replaced) [pdf, html, other]
-
Title: Decentralized Low-Rank Fine-Tuning of Large Language ModelsSubjects: Machine Learning (cs.LG)
While parameter-efficient fine-tuning (PEFT) techniques like Low-Rank Adaptation (LoRA) offer computationally efficient adaptations of Large Language Models (LLMs), their practical deployment often assumes centralized data and training environments. However, real-world scenarios frequently involve distributed, privacy-sensitive datasets that require decentralized solutions. Federated learning (FL) addresses data privacy by coordinating model updates across clients, but it is typically based on centralized aggregation through a parameter server, which can introduce bottlenecks and communication constraints. Decentralized learning, in contrast, eliminates this dependency by enabling direct collaboration between clients, improving scalability and efficiency in distributed environments. Despite its advantages, decentralized LLM fine-tuning remains underexplored. In this work, we propose Dec-LoRA, a decentralized fine-tuning algorithm for LLMs based on LoRA. Through extensive experiments on BERT and LLaMA-2 models, we demonstrate that Dec-LoRA achieves performance comparable to centralized LoRA under various conditions, including data heterogeneity and quantization constraints. Additionally, we provide a rigorous theoretical guarantee proving the convergence of our algorithm to a stationary point for non-convex and smooth loss functions. These findings highlight the potential of Dec-LoRA for scalable LLM fine-tuning in decentralized environments.
- [493] arXiv:2501.16398 (replaced) [pdf, other]
-
Title: Data-Efficient Machine Learning Potentials via Difference Vectors Based on Local Atomic EnvironmentsXuqiang Shao, Yuqi Zhang, Di Zhang, Zhaoyan Dong, Tianxiang Gao, Mingzhe Li, Xinyuan Liu, Zhiran Gan, Fanshun Meng, Lingcai Kong, Zhengyang Gao, Hao Lic, Weijie YangdSubjects: Machine Learning (cs.LG); Atomic Physics (physics.atom-ph)
Constructing efficient and diverse datasets is essential for the development of accurate machine learning potentials (MLPs) in atomistic simulations. However, existing approaches often suffer from data redundancy and high computational costs. Herein, we propose a new method--Difference Vectors based on Local Atomic Environments (DV-LAE)--that encodes structural differences via histogram-based descriptors and enables visual analysis through t-SNE dimensionality reduction. This approach facilitates redundancy detection and dataset optimization while preserving structural diversity. We demonstrate that DV-LAE significantly reduces dataset size and training time across various materials systems, including high-pressure hydrogen, iron-hydrogen binaries, magnesium hydrides, and carbon allotropes, with minimal compromise in prediction accuracy. For instance, in the $\alpha$-Fe/H system, maintaining a highly similar MLP accuracy, the dataset size was reduced by 56%, and the training time per iteration dropped by over 50%. Moreover, we show how visualizing the DV-LAE representation aids in identifying out-of-distribution data by examining the spatial distribution of high-error prediction points, providing a robust reliability metric for new structures during simulations. Our results highlight the utility of local environment visualization not only as an interpretability tool but also as a practical means for accelerating MLP development and ensuring data efficiency in large-scale atomistic modeling.
- [494] arXiv:2501.17077 (replaced) [pdf, html, other]
-
Title: Inducing, Detecting and Characterising Neural Modules: A Pipeline for Functional Interpretability in Reinforcement LearningSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Interpretability is crucial for ensuring RL systems align with human values. However, it remains challenging to achieve in complex decision making domains. Existing methods frequently attempt interpretability at the level of fundamental model units, such as neurons or decision nodes: an approach which scales poorly to large models. Here, we instead propose an approach to interpretability at the level of functional modularity. We show how encouraging sparsity and locality in network weights leads to the emergence of functional modules in RL policy networks. To detect these modules, we develop an extended Louvain algorithm which uses a novel `correlation alignment' metric to overcome the limitations of standard network analysis techniques when applied to neural network architectures. Applying these methods to 2D and 3D MiniGrid environments reveals the consistent emergence of distinct navigational modules for different axes, and we further demonstrate how these functions can be validated through direct interventions on network weights prior to inference.
- [495] arXiv:2501.18092 (replaced) [pdf, html, other]
-
Title: Learning Provably Improves the Convergence of Gradient DescentComments: 45 pages, 10 figuresSubjects: Machine Learning (cs.LG); Optimization and Control (math.OC)
Learn to Optimize (L2O) trains deep neural network based solvers for optimization, achieving success in accelerating convex problems and improving non-convex solutions. However, L2O lacks rigorous theoretical backing for its own training convergence, as existing analyses often use unrealistic assumptions -- a gap this work highlights empirically. We bridge this gap by proving the training convergence of L2O models that learn Gradient Descent (GD) hyperparameters for quadratic programming, leveraging the Neural Tangent Kernel (NTK) theory. We propose a deterministic initialization strategy to support our theoretical results and promote stable training over extended optimization horizons by mitigating gradient explosion. Our L2O framework demonstrates over 50\% better optimality against GD and superior robustness over state-of-the-art L2O methods on synthetic datasets.
- [496] arXiv:2501.18915 (replaced) [pdf, html, other]
-
Title: Algebra Unveils Deep Learning -- An Invitation to Neuroalgebraic GeometryComments: Published at ICML 2025Subjects: Machine Learning (cs.LG); Algebraic Geometry (math.AG)
In this position paper, we promote the study of function spaces parameterized by machine learning models through the lens of algebraic geometry. To this end, we focus on algebraic models, such as neural networks with polynomial activations, whose associated function spaces are semi-algebraic varieties. We outline a dictionary between algebro-geometric invariants of these varieties, such as dimension, degree, and singularities, and fundamental aspects of machine learning, such as sample complexity, expressivity, training dynamics, and implicit bias. Along the way, we review the literature and discuss ideas beyond the algebraic domain. This work lays the foundations of a research direction bridging algebraic geometry and deep learning, that we refer to as neuroalgebraic geometry.
- [497] arXiv:2501.18935 (replaced) [pdf, other]
-
Title: TabFSBench: Tabular Benchmark for Feature Shifts in Open EnvironmentsSubjects: Machine Learning (cs.LG)
Tabular data is widely utilized in various machine learning tasks. Current tabular learning research predominantly focuses on closed environments, while in real-world applications, open environments are often encountered, where distribution and feature shifts occur, leading to significant degradation in model performance. Previous research has primarily concentrated on mitigating distribution shifts, whereas feature shifts, a distinctive and unexplored challenge of tabular data, have garnered limited attention. To this end, this paper conducts the first comprehensive study on feature shifts in tabular data and introduces the first tabular feature-shift benchmark (TabFSBench). TabFSBench evaluates impacts of four distinct feature-shift scenarios on four tabular model categories across various datasets and assesses the performance of large language models (LLMs) and tabular LLMs in the tabular benchmark for the first time. Our study demonstrates three main observations: (1) most tabular models have the limited applicability in feature-shift scenarios; (2) the shifted feature set importance has a linear relationship with model performance degradation; (3) model performance in closed environments correlates with feature-shift performance. Future research direction is also explored for each observation.
Benchmark: this https URL. - [498] arXiv:2501.19107 (replaced) [pdf, html, other]
-
Title: Brain network science modelling of sparse neural networks enables Transformers and LLMs to perform as fully connectedYingtao Zhang, Diego Cerretti, Jialin Zhao, Wenjing Wu, Ziheng Liao, Umberto Michieli, Carlo Vittorio CannistraciSubjects: Machine Learning (cs.LG)
Dynamic sparse training (DST) can reduce the computational demands in ANNs, but faces difficulties in keeping peak performance at high sparsity levels. The Cannistraci-Hebb training (CHT) is a brain-inspired method for growing connectivity in DST. CHT leverages a gradient-free, topology-driven link regrowth, which has shown ultra-sparse (less than 1% connectivity) advantage across various tasks compared to fully connected networks. Yet, CHT suffers two main drawbacks: (i) its time complexity is $O(Nd^3)$ - N node network size, d node degree - restricting it to ultra-sparse regimes. (ii) it selects top link prediction scores, which is inappropriate for the early training epochs, when the network presents unreliable connections. Here, we design the first brain-inspired network model - termed bipartite receptive field (BRF) - to initialize the connectivity of sparse artificial neural networks. We further introduce a GPU-friendly matrix-based approximation of CH link prediction, reducing complexity to $O(N^3)$. We introduce the Cannistraci-Hebb training soft rule (CHTs), which adopts a flexible strategy for sampling connections in both link removal and regrowth, balancing the exploration and exploitation of network topology. Additionally, we integrate CHTs with a sigmoid gradual density decay (CHTss). Empirical results show that BRF offers performance advantages over previous network science models. Using 1% of connections, CHTs outperforms fully connected networks in MLP architectures on image classification tasks, compressing some networks to less than 30% of the nodes. Using 5% of the connections, CHTss outperforms fully connected networks in two Transformer-based machine translation tasks. Finally, at 30% connectivity, both CHTs and CHTss outperform other DST methods in language modeling and even exceed fully connected baselines in zero-shot tasks.
- [499] arXiv:2501.19358 (replaced) [pdf, html, other]
-
Title: The Energy Loss Phenomenon in RLHF: A New Perspective on Mitigating Reward HackingComments: The paper has been accepted by ICML 2025Subjects: Machine Learning (cs.LG)
This work identifies the Energy Loss Phenomenon in Reinforcement Learning from Human Feedback (RLHF) and its connection to reward hacking. Specifically, energy loss in the final layer of a Large Language Model (LLM) gradually increases during the RL process, with an excessive increase in energy loss characterizing reward hacking. Beyond empirical analysis, we further provide a theoretical foundation by proving that, under mild conditions, the increased energy loss reduces the upper bound of contextual relevance in LLMs, which is a critical aspect of reward hacking as the reduced contextual relevance typically indicates overfitting to reward model-favored patterns in RL. To address this issue, we propose an Energy loss-aware PPO algorithm (EPPO) which penalizes the increase in energy loss in the LLM's final layer during reward calculation to prevent excessive energy loss, thereby mitigating reward hacking. We theoretically show that EPPO can be conceptually interpreted as an entropy-regularized RL algorithm, which provides deeper insights into its effectiveness. Extensive experiments across various LLMs and tasks demonstrate the commonality of the energy loss phenomenon, as well as the effectiveness of EPPO in mitigating reward hacking and improving RLHF performance.
- [500] arXiv:2502.00180 (replaced) [pdf, other]
-
Title: Spectral Analysis of Diffusion Models with Application to Schedule DesignSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Diffusion models (DMs) have emerged as powerful tools for modeling complex data distributions and generating realistic new samples. Over the years, advanced architectures and sampling methods have been developed to make these models practically usable. However, certain synthesis process decisions still rely on heuristics without a solid theoretical foundation. In our work, we offer a novel analysis of the DM's inference process, introducing a comprehensive frequency response perspective. Specifically, by relying on Gaussianity assumption, we present the inference process as a closed-form spectral transfer function, capturing how the generated signal evolves in response to the initial noise. We demonstrate how the proposed analysis can be leveraged to design a noise schedule that aligns effectively with the characteristics of the data. The spectral perspective also provides insights into the underlying dynamics and sheds light on the relationship between spectral properties and noise schedule structure. Our results lead to scheduling curves that are dependent on the spectral content of the data, offering a theoretical justification for some of the heuristics taken by practitioners.
- [501] arXiv:2502.00407 (replaced) [pdf, html, other]
-
Title: Causal Abstraction Learning based on the Semantic Embedding PrincipleJournal-ref: Proceedings of the Forty-second International Conference on Machine Learning (ICML 2025)Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Structural causal models (SCMs) allow us to investigate complex systems at multiple levels of resolution. The causal abstraction (CA) framework formalizes the mapping between high- and low-level SCMs. We address CA learning in a challenging and realistic setting, where SCMs are inaccessible, interventional data is unavailable, and sample data is misaligned. A key principle of our framework is semantic embedding, formalized as the high-level distribution lying on a subspace of the low-level one. This principle naturally links linear CA to the geometry of the Stiefel manifold. We present a category-theoretic approach to SCMs that enables the learning of a CA by finding a morphism between the low- and high-level probability measures, adhering to the semantic embedding principle. Consequently, we formulate a general CA learning problem. As an application, we solve the latter problem for linear CA; considering Gaussian measures and the Kullback-Leibler divergence as an objective. Given the nonconvexity of the learning task, we develop three algorithms building upon existing paradigms for Riemannian optimization. We demonstrate that the proposed methods succeed on both synthetic and real-world brain data with different degrees of prior information about the structure of CA.
- [502] arXiv:2502.00463 (replaced) [pdf, html, other]
-
Title: Efficient Over-parameterized Matrix Sensing from Noisy Measurements via Alternating Preconditioned Gradient DescentComments: 18 pages, 8 figuresSubjects: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
We consider the noisy matrix sensing problem in the over-parameterization setting, where the estimated rank $r$ is larger than the true rank $r_\star$ of the target matrix $X_\star$. Specifically, our main objective is to recover a matrix $ X_\star \in \mathbb{R}^{n_1 \times n_2} $ with rank $ r_\star $ from noisy measurements using an over-parameterized factorization $ LR^\top $, where $ L \in \mathbb{R}^{n_1 \times r}, \, R \in \mathbb{R}^{n_2 \times r} $ and $ \min\{n_1, n_2\} \ge r > r_\star $, with $ r_\star $ being unknown. Recently, preconditioning methods have been proposed to accelerate the convergence of matrix sensing problem compared to vanilla gradient descent, incorporating preconditioning terms $ (L^\top L + \lambda I)^{-1} $ and $ (R^\top R + \lambda I)^{-1} $ into the original gradient. However, these methods require careful tuning of the damping parameter $\lambda$ and are sensitive to step size. To address these limitations, we propose the alternating preconditioned gradient descent (APGD) algorithm, which alternately updates the two factor matrices, eliminating the need for the damping parameter $\lambda$ and enabling faster convergence with larger step sizes. We theoretically prove that APGD convergences to a near-optimal error at a linear rate. We further show that APGD can be extended to deal with other low-rank matrix estimation tasks, also with a theoretical guarantee of linear convergence. To validate the effectiveness and scalability of the proposed APGD, we conduct simulated and real-world experiments on a wide range of low-rank estimation problems, including noisy matrix sensing, weighted PCA, 1-bit matrix completion, and matrix completion. The extensive results demonstrate that APGD consistently achieves the fastest convergence and the lowest computation time compared to the existing alternatives.
- [503] arXiv:2502.00657 (replaced) [pdf, html, other]
-
Title: LLM Safety Alignment is Divergence Estimation in DisguiseSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Machine Learning (stat.ML)
We present a theoretical framework showing that popular LLM alignment methods, including RLHF and its variants, can be understood as divergence estimators between aligned (safe or preferred) and unaligned (harmful or less preferred) distributions. This perspective explains the emergence of separation in the latent space between safe and harmful prompts after alignment. As an application of our general divergence framework, we propose KLDO, a novel KL divergence-based alignment method, and empirically validate its effectiveness. We further show that using compliance-refusal datasets, rather than standard preference-based datasets, leads to stronger separation and improved safety alignment. Finally, to quantify the separation effect, we propose a distance-based metric in the prompt representation space, which also acts as a statistically significant indicator for model safety.
- [504] arXiv:2502.01159 (replaced) [pdf, other]
-
Title: AtmosSci-Bench: Evaluating the Recent Advance of Large Language Model for Atmospheric ScienceComments: 33 pages, 4 figures, 7 tablesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
The rapid advancements in large language models (LLMs), particularly in their reasoning capabilities, hold transformative potential for addressing complex challenges in atmospheric science. However, leveraging LLMs effectively in this domain requires a robust and comprehensive evaluation benchmark. Toward this end, we present AtmosSci-Bench, a novel benchmark designed to systematically assess LLM performance across five core categories of atmospheric science problems: hydrology, atmospheric dynamics, atmospheric physics, geophysics, and physical oceanography. AtmosSci-Bench features a dual-format design comprising both multiple-choice questions (MCQs) and open-ended questions (OEQs), enabling scalable automated evaluation alongside deeper analysis of conceptual understanding. We employ a template-based MCQ generation framework to create diverse, graduate-level problems with symbolic perturbation, while OEQs are used to probe open-ended reasoning. We conduct a comprehensive evaluation of representative LLMs, categorized into four groups: instruction-tuned models, advanced reasoning models, math-augmented models, and domain-specific climate models. Our analysis provides some interesting insights into the reasoning and problem-solving capabilities of LLMs in atmospheric science. We believe AtmosSci-Bench can serve as a critical step toward advancing LLM applications in climate service by offering a standard and rigorous evaluation framework. Our source codes are currently available at Our source codes are currently available at this https URL.
- [505] arXiv:2502.01591 (replaced) [pdf, html, other]
-
Title: Improving Transformer World Models for Data-Efficient RLAntoine Dedieu, Joseph Ortiz, Xinghua Lou, Carter Wendelken, Wolfgang Lehrach, J Swaroop Guntupalli, Miguel Lazaro-Gredilla, Kevin Patrick MurphyComments: ICML 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
We present an approach to model-based RL that achieves a new state of the art performance on the challenging Craftax-classic benchmark, an open-world 2D survival game that requires agents to exhibit a wide range of general abilities -- such as strong generalization, deep exploration, and long-term reasoning. With a series of careful design choices aimed at improving sample efficiency, our MBRL algorithm achieves a reward of 69.66% after only 1M environment steps, significantly outperforming DreamerV3, which achieves 53.2%, and, for the first time, exceeds human performance of 65.0%. Our method starts by constructing a SOTA model-free baseline, using a novel policy architecture that combines CNNs and RNNs. We then add three improvements to the standard MBRL setup: (a) "Dyna with warmup", which trains the policy on real and imaginary data, (b) "nearest neighbor tokenizer" on image patches, which improves the scheme to create the transformer world model (TWM) inputs, and (c) "block teacher forcing", which allows the TWM to reason jointly about the future tokens of the next timestep.
- [506] arXiv:2502.01989 (replaced) [pdf, html, other]
-
Title: VFScale: Intrinsic Reasoning through Verifier-Free Test-time Scalable Diffusion ModelComments: 22 pages, 13 figuresSubjects: Machine Learning (cs.LG)
Inspired by human SYSTEM 2 thinking, LLMs excel at complex reasoning tasks via extended Chain-of-Thought. However, similar test-time scaling for diffusion models to tackle complex reasoning remains largely unexplored. From existing work, two primary challenges emerge in this setting: (i) the dependence on an external verifier indicating a notable gap from intrinsic reasoning of human intelligence without any external feedback, and (ii) the lack of an efficient search algorithm. In this paper, we introduce the Verifier-free Test-time Scalable Diffusion Model (VFScale) to achieve scalable intrinsic reasoning, which equips number-of-sample test-time scaling with the intrinsic energy function of diffusion models as the verifier. Concretely, VFScale comprises two key innovations to address the aforementioned challenges. On the training side, VFScale consists of a novel LRNCL loss and a KL regularization to improve the energy landscape, ensuring that the learned energy function itself serves as a reliable verifier. On the inference side, VFScale integrates the denoising process with a novel hybrid Monte Carlo Tree Search (hMCTS) to improve search efficiency. On challenging reasoning tasks of Maze and Sudoku, we demonstrate the effectiveness of VFScale's training objective and scalable inference method. In particular, trained with Maze sizes of up to $6\times6$, our VFScale solves 88% of Maze problems with much larger sizes of $15\times15$, while standard diffusion model completely fails.
- [507] arXiv:2502.04034 (replaced) [pdf, html, other]
-
Title: Fourier Asymmetric Attention on Domain Generalization for Pan-Cancer Drug Response PredictionSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
The accurate prediction of drug responses remains a formidable challenge, particularly at the single-cell level and in clinical treatment contexts. Some studies employ transfer learning techniques to predict drug responses in individual cells and patients, but they require access to target-domain data during training, which is often unavailable or only obtainable in future. In this study, we propose a novel domain generalization framework, termed FourierDrug, to address this challenge. Given the extracted feature from expression profile, we performed Fourier transforms and then introduced an asymmetric attention constraint that would cluster drug-sensitive samples into a compact group while drives resistant samples dispersed in the frequency domain. Our empirical experiments demonstrate that our model effectively learns task-relevant features from diverse source domains, and achieves accurate predictions of drug response for unseen cancer type. When evaluated on single-cell and patient-level drug response prediction tasks, FourierDrug--trained solely on in vitro cell line data without access to target-domain data--consistently outperforms or, at least, matched the performance of current state-of-the-art methods. These findings underscore the potential of our method for real-world clinical applications.
- [508] arXiv:2502.04420 (replaced) [pdf, other]
-
Title: KVTuner: Sensitivity-Aware Layer-Wise Mixed-Precision KV Cache Quantization for Efficient and Nearly Lossless LLM InferenceXing Li, Zeyu Xing, Yiming Li, Linping Qu, Hui-Ling Zhen, Wulong Liu, Yiwu Yao, Sinno Jialin Pan, Mingxuan YuanComments: Accepted by ICML25. Code: this https URLSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
KV cache quantization can improve Large Language Models (LLMs) inference throughput and latency in long contexts and large batch-size scenarios while preserving LLMs effectiveness. However, current methods have three unsolved issues: overlooking layer-wise sensitivity to KV cache quantization, high overhead of online fine-grained decision-making, and low flexibility to different LLMs and constraints. Therefore, we theoretically analyze the inherent correlation of layer-wise transformer attention patterns to KV cache quantization errors and study why key cache is generally more important than value cache for quantization error reduction. We further propose a simple yet effective framework KVTuner to adaptively search for the optimal hardware-friendly layer-wise KV quantization precision pairs for coarse-grained KV cache with multi-objective optimization and directly utilize the offline searched configurations during online inference. To reduce the computational cost of offline calibration, we utilize the intra-layer KV precision pair pruning and inter-layer clustering to reduce the search space. Experimental results show that we can achieve nearly lossless 3.25-bit mixed precision KV cache quantization for LLMs like Llama-3.1-8B-Instruct and 4.0-bit for sensitive models like Qwen2.5-7B-Instruct on mathematical reasoning tasks. The maximum inference throughput can be improved by 21.25\% compared with KIVI-KV8 quantization over various context lengths. Our code and searched configurations are available at this https URL.
- [509] arXiv:2502.04809 (replaced) [pdf, html, other]
-
Title: Humans Coexist, So Must Embodied Artificial AgentsSubjects: Machine Learning (cs.LG)
This paper introduces the concept of coexistence for embodied artificial agents and argues that it is a prerequisite for long-term, in-the-wild interaction with humans. Contemporary embodied artificial agents excel in static, predefined tasks but fall short in dynamic and long-term interactions with humans. On the other hand, humans can adapt and evolve continuously, exploiting the situated knowledge embedded in their environment and other agents, thus contributing to meaningful interactions. We take an interdisciplinary approach at different levels of organization, drawing from biology and design theory, to understand how human and non-human organisms foster entities that coexist within their specific environments. Finally, we propose key research directions for the artificial intelligence community to develop coexisting embodied agents, focusing on the principles, hardware and learning methods responsible for shaping them.
- [510] arXiv:2502.06051 (replaced) [pdf, html, other]
-
Title: Towards a Sharp Analysis of Offline Policy Learning for $f$-Divergence-Regularized Contextual BanditsComments: 38 pagesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Statistics Theory (math.ST); Machine Learning (stat.ML)
Although many popular reinforcement learning algorithms are underpinned by $f$-divergence regularization, their sample complexity with respect to the \emph{regularized objective} still lacks a tight characterization. In this paper, we analyze $f$-divergence-regularized offline policy learning. For reverse Kullback-Leibler (KL) divergence, arguably the most commonly used one, we give the first $\tilde{O}(\epsilon^{-1})$ sample complexity under single-policy concentrability for contextual bandits, surpassing existing $\tilde{O}(\epsilon^{-1})$ bound under all-policy concentrability and $\tilde{O}(\epsilon^{-2})$ bound under single-policy concentrability. Our analysis for general function approximation leverages the principle of pessimism in the face of uncertainty to refine a mean-value-type argument to its extreme. This in turn leads to a novel moment-based technique, effectively bypassing the need for uniform control over the discrepancy between any two functions in the function class. We further propose a lower bound, demonstrating that a multiplicative dependency on single-policy concentrability is necessary to maximally exploit the strong convexity of reverse KL. In addition, for $f$-divergences with strongly convex $f$, to which reverse KL \emph{does not} belong, we show that the sharp sample complexity $\tilde{\Theta}(\epsilon^{-1})$ is achievable even without single-policy concentrability. In this case, the algorithm design can get rid of pessimistic estimators. We further extend our analysis to dueling bandits, and we believe these results take a significant step toward a comprehensive understanding of $f$-divergence-regularized policy learning.
- [511] arXiv:2502.06751 (replaced) [pdf, html, other]
-
Title: What makes a good feedforward computational graph?Comments: To appear at ICML 2025. 17 pages, 7 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Social and Information Networks (cs.SI); Machine Learning (stat.ML)
As implied by the plethora of literature on graph rewiring, the choice of computational graph employed by a neural network can make a significant impact on its downstream performance. Certain effects related to the computational graph, such as under-reaching and over-squashing, may even render the model incapable of learning certain functions. Most of these effects have only been thoroughly studied in the domain of undirected graphs; however, recent years have seen a significant rise in interest in feedforward computational graphs: directed graphs without any back edges. In this paper, we study the desirable properties of a feedforward computational graph, discovering two important complementary measures: fidelity and mixing time, and evaluating a few popular choices of graphs through the lens of these measures. Our study is backed by both theoretical analyses of the metrics' asymptotic behaviour for various graphs, as well as correlating these metrics to the performance of trained neural network models using the corresponding graphs.
- [512] arXiv:2502.07460 (replaced) [pdf, html, other]
-
Title: Logarithmic Regret for Online KL-Regularized Reinforcement LearningSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Recent advances in Reinforcement Learning from Human Feedback (RLHF) have shown that KL-regularization plays a pivotal role in improving the efficiency of RL fine-tuning for large language models (LLMs). Despite its empirical advantage, the theoretical difference between KL-regularized RL and standard RL remains largely under-explored. While there is a recent line of work on the theoretical analysis of KL-regularized objective in decision making \citep{xiong2024iterative, xie2024exploratory,zhao2024sharp}, these analyses either reduce to the traditional RL setting or rely on strong coverage assumptions. In this paper, we propose an optimism-based KL-regularized online contextual bandit algorithm, and provide a novel analysis of its regret. By carefully leveraging the benign optimization landscape induced by the KL-regularization and the optimistic reward estimation, our algorithm achieves an $\mathcal{O}\big(\eta\log (N_{\mathcal R} T)\cdot d_{\mathcal R}\big)$ logarithmic regret bound, where $\eta, N_{\mathcal R},T,d_{\mathcal R}$ denote the KL-regularization parameter, the cardinality of the reward function class, number of rounds, and the complexity of the reward function class. Furthermore, we extend our algorithm and analysis to reinforcement learning by developing a novel decomposition over transition steps and also obtain a similar logarithmic regret bound.
- [513] arXiv:2502.08202 (replaced) [pdf, html, other]
-
Title: Privacy amplification by random allocationSubjects: Machine Learning (cs.LG)
We consider the privacy amplification properties of a sampling scheme in which a user's data is used in $k$ steps chosen randomly and uniformly from a sequence (or set) of $t$ steps. This sampling scheme has been recently applied in the context of differentially private optimization [Chua et al., 2024a, Choquette-Choo et al., 2024] and is also motivated by communication-efficient high-dimensional private aggregation [Asi et al., 2025]. Existing analyses of this scheme either rely on privacy amplification by shuffling which leads to overly conservative bounds or require Monte Carlo simulations that are computationally prohibitive in most practical scenarios.
We give the first theoretical guarantees and numerical estimation algorithms for this sampling scheme. In particular, we demonstrate that the privacy guarantees of random $k$-out-of-$t$ allocation can be upper bounded by the privacy guarantees of the well-studied independent (or Poisson) subsampling in which each step uses the user's data with probability $(1+o(1))k/t$. Further, we provide two additional analysis techniques that lead to numerical improvements in several parameter regimes. Altogether, our bounds give efficiently-computable and nearly tight numerical results for random allocation applied to Gaussian noise addition. - [514] arXiv:2502.08598 (replaced) [pdf, html, other]
-
Title: Disentangling Total-Variance and Signal-to-Noise-Ratio Improves Diffusion ModelsKhaled Kahouli, Winfried Ripken, Stefan Gugler, Oliver T. Unke, Klaus-Robert Müller, Shinichi NakajimaSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
The long sampling time of diffusion models remains a significant bottleneck, which can be mitigated by reducing the number of diffusion time steps. However, the quality of samples with fewer steps is highly dependent on the noise schedule, i.e., the specific manner in which noise is introduced and the signal is reduced at each step. Although prior work has improved upon the original variance-preserving and variance-exploding schedules, these approaches $\textit{passively}$ adjust the total variance, without direct control over it. In this work, we propose a novel total-variance/signal-to-noise-ratio disentangled (TV/SNR) framework, where TV and SNR can be controlled independently. Our approach reveals that schedules where the TV explodes exponentially can often be improved by adopting a constant TV schedule while preserving the same SNR schedule. Furthermore, generalizing the SNR schedule of the optimal transport flow matching significantly improves the generation performance. Our findings hold across various reverse diffusion solvers and diverse applications, including molecular structure and image generation.
- [515] arXiv:2502.08942 (replaced) [pdf, html, other]
-
Title: Language in the Flow of Time: Time-Series-Paired Texts Weaved into a Unified Temporal NarrativeZihao Li, Xiao Lin, Zhining Liu, Jiaru Zou, Ziwei Wu, Lecheng Zheng, Dongqi Fu, Yada Zhu, Hendrik Hamann, Hanghang Tong, Jingrui HeComments: Preprint, 43 pagesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
While many advances in time series models focus exclusively on numerical data, research on multimodal time series, particularly those involving contextual textual information commonly encountered in real-world scenarios, remains in its infancy. With recent progress in large language models and time series learning, we revisit the integration of paired texts with time series through the Platonic Representation Hypothesis, which posits that representations of different modalities converge to shared spaces. In this context, we identify that time-series-paired texts may naturally exhibit periodic properties that closely mirror those of the original time series. Building on this insight, we propose a novel framework, Texts as Time Series (TaTS), which considers the time-series-paired texts to be auxiliary variables of the time series. TaTS can be plugged into any existing numerical-only time series models and enable them to handle time series data with paired texts effectively. Through extensive experiments on both multimodal time series forecasting and imputation tasks across benchmark datasets with various existing time series models, we demonstrate that TaTS can enhance predictive performance without modifying model architectures. Code available at this https URL.
- [516] arXiv:2502.09254 (replaced) [pdf, html, other]
-
Title: AnomalyGFM: Graph Foundation Model for Zero/Few-shot Anomaly DetectionComments: Accepted by KDD2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Graph anomaly detection (GAD) aims to identify abnormal nodes that differ from the majority of the nodes in a graph, which has been attracting significant attention in recent years. Existing generalist graph models have achieved remarkable success in different graph tasks but struggle to generalize to the GAD task. This limitation arises from their difficulty in learning generalized knowledge for capturing the inherently infrequent, irregular and heterogeneous abnormality patterns in graphs from different domains. To address this challenge, we propose AnomalyGFM, a GAD-oriented graph foundation model that supports zero-shot inference and few-shot prompt tuning for GAD in diverse graph datasets. One key insight is that graph-agnostic representations for normal and abnormal classes are required to support effective zero/few-shot GAD across different graphs. Motivated by this, AnomalyGFM is pre-trained to align data-independent, learnable normal and abnormal class prototypes with node representation residuals (i.e., representation deviation of a node from its neighbors). The residual features essentially project the node information into a unified feature space where we can effectively measure the abnormality of nodes from different graphs in a consistent way. This provides a driving force for the learning of graph-agnostic, discriminative prototypes for the normal and abnormal classes, which can be used to enable zero-shot GAD on new graphs, including very large-scale graphs. If there are few-shot labeled normal nodes available in the new graphs, AnomalyGFM can further support prompt tuning to leverage these nodes for better adaptation. Comprehensive experiments on 11 widely-used GAD datasets with real anomalies, demonstrate that AnomalyGFM significantly outperforms state-of-the-art competing methods under both zero- and few-shot GAD settings.
- [517] arXiv:2502.09365 (replaced) [pdf, html, other]
-
Title: Simple Path Structural Encoding for Graph TransformersSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Graph transformers extend global self-attention to graph-structured data, achieving notable success in graph learning. Recently, random walk structural encoding (RWSE) has been found to further enhance their predictive power by encoding both structural and positional information into the edge representation. However, RWSE cannot always distinguish between edges that belong to different local graph patterns, which reduces its ability to capture the full structural complexity of graphs. This work introduces Simple Path Structural Encoding (SPSE), a novel method that utilizes simple path counts for edge encoding. We show theoretically and experimentally that SPSE overcomes the limitations of RWSE, providing a richer representation of graph structures, particularly for capturing local cyclic patterns. To make SPSE computationally tractable, we propose an efficient approximate algorithm for simple path counting. SPSE demonstrates significant performance improvements over RWSE on various benchmarks, including molecular and long-range graph datasets, achieving statistically significant gains in discriminative tasks. These results pose SPSE as a powerful edge encoding alternative for enhancing the expressivity of graph transformers.
- [518] arXiv:2502.10762 (replaced) [pdf, html, other]
-
Title: Bone Soups: A Seek-and-Soup Model Merging Approach for Controllable Multi-Objective GenerationComments: This paper is accepted by the ACL 2025 Main ConferenceSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
User information needs are often highly diverse and varied. A key challenge in current research is how to achieve controllable multi-objective generation while enabling rapid adaptation to accommodate diverse user demands during test time. Existing solutions, such as Rewarded Soup, focus on merging language models individually tuned on single objectives. While easy to implement and widely used, these approaches face limitations in achieving optimal performance due to their disregard for the impacts of competing objectives on model tuning. To address this issue, we propose Bone Soup, a novel model merging approach that first seeks a series of backbone models by considering the impacts of multiple objectives and then makes the soup (i.e., merge the backbone models). Specifically, Bone Soup begins by training multiple backbone models for different objectives using multi-objective reinforcement learning. Each backbone model is guided by a combination of backbone reward signals. To ensure that these models are optimal for the Pareto front, the backbone rewards are crafted by combining standard reward functions into basis vectors, which can then be modified through a rule-based construction method. Bone Soup leverages a symmetric circulant matrix mapping to generate the merging coefficients, which are used to merge the backbone models according to user preferences. Extensive experimental results demonstrate that Bone Soup exhibits strong controllability and Pareto optimality in controllable multi-objective generation, providing a more effective and efficient approach to addressing diverse user needs at test time.
- [519] arXiv:2502.11196 (replaced) [pdf, html, other]
-
Title: How Do LLMs Acquire New Knowledge? A Knowledge Circuits Perspective on Continual Pre-TrainingComments: ACL 2025 FindingsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC)
Despite exceptional capabilities in knowledge-intensive tasks, Large Language Models (LLMs) face a critical gap in understanding how they internalize new knowledge, particularly how to structurally embed acquired knowledge in their neural computations. We address this issue through the lens of knowledge circuit evolution, identifying computational subgraphs that facilitate knowledge storage and processing. Our systematic analysis of circuit evolution throughout continual pre-training reveals several key findings: (1) the acquisition of new knowledge is influenced by its relevance to pre-existing knowledge; (2) the evolution of knowledge circuits exhibits a distinct phase shift from formation to optimization; (3) the evolution of knowledge circuits follows a deep-to-shallow pattern. These insights not only advance our theoretical understanding of the mechanisms of new knowledge acquisition in LLMs, but also provide potential implications for improving continual pre-training strategies to enhance model performance. Code and data will be available at this https URL.
- [520] arXiv:2502.11767 (replaced) [pdf, html, other]
-
Title: From Selection to Generation: A Survey of LLM-based Active LearningYu Xia, Subhojyoti Mukherjee, Zhouhang Xie, Junda Wu, Xintong Li, Ryan Aponte, Hanjia Lyu, Joe Barrow, Hongjie Chen, Franck Dernoncourt, Branislav Kveton, Tong Yu, Ruiyi Zhang, Jiuxiang Gu, Nesreen K. Ahmed, Yu Wang, Xiang Chen, Hanieh Deilamsalehy, Sungchul Kim, Zhengmian Hu, Yue Zhao, Nedim Lipka, Seunghyun Yoon, Ting-Hao Kenneth Huang, Zichao Wang, Puneet Mathur, Soumyabrata Pal, Koyel Mukherjee, Zhehao Zhang, Namyong Park, Thien Huu Nguyen, Jiebo Luo, Ryan A. Rossi, Julian McAuleyComments: ACL 2025Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Active Learning (AL) has been a powerful paradigm for improving model efficiency and performance by selecting the most informative data points for labeling and training. In recent active learning frameworks, Large Language Models (LLMs) have been employed not only for selection but also for generating entirely new data instances and providing more cost-effective annotations. Motivated by the increasing importance of high-quality data and efficient model training in the era of LLMs, we present a comprehensive survey on LLM-based Active Learning. We introduce an intuitive taxonomy that categorizes these techniques and discuss the transformative roles LLMs can play in the active learning loop. We further examine the impact of AL on LLM learning paradigms and its applications across various domains. Finally, we identify open challenges and propose future research directions. This survey aims to serve as an up-to-date resource for researchers and practitioners seeking to gain an intuitive understanding of LLM-based AL techniques and deploy them to new applications.
- [521] arXiv:2502.13576 (replaced) [pdf, html, other]
-
Title: Beyond One-Size-Fits-All: Tailored Benchmarks for Efficient EvaluationPeiwen Yuan, Yueqi Zhang, Shaoxiong Feng, Yiwei Li, Xinglin Wang, Jiayi Shi, Chuyi Tan, Boyuan Pan, Yao Hu, Kan LiSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Evaluating models on large benchmarks is very resource-intensive, especially during the period of rapid model evolution. Existing efficient evaluation methods estimate the performance of target models by testing them only on a small and static coreset of the benchmark, which is derived from the publicly available evaluation results of source models. These methods rely on the assumption that target models have high prediction consistency with source models. However, we demonstrate that it doesn't generalize well in practice. To alleviate the inconsistency issue, we present TailoredBench, a method that conducts customized evaluation tailored to each target model. Specifically, a Global-coreset is first constructed as a probe to identify the most consistent source models for each target model with an adaptive source model selection strategy. Afterwards, a scalable K-Medoids clustering algorithm is proposed to extend the Global-coreset to a tailored Native-coreset for each target model. According to the predictions on Native-coresets, we obtain the performance of target models on the whole benchmark with a calibrated estimation strategy. Comprehensive experiments on 5 benchmarks across over 300 models demonstrate that compared to best performing baselines, TailoredBench achieves an average reduction of 31.4% in MAE of accuracy estimates under the same inference budgets, showcasing strong effectiveness and generalizability.
- [522] arXiv:2502.15296 (replaced) [pdf, other]
-
Title: Beyond Fixed Variables: Expanding-variate Time Series Forecasting via Flat Scheme and Spatio-temporal Focal LearningSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Multivariate Time Series Forecasting (MTSF) has long been a key research focus. Traditionally, these studies assume a fixed number of variables, but in real-world applications, Cyber-Physical Systems often expand as new sensors are deployed, increasing variables in MTSF. In light of this, we introduce a novel task, Expanding-variate Time Series Forecasting (EVTSF). This task presents unique challenges, specifically (1) handling inconsistent data shapes caused by adding new variables, and (2) addressing imbalanced spatio-temporal learning, where expanding variables have limited observed data due to the necessity for timely operation. To address these challenges, we propose STEV, a flexible spatio-temporal forecasting framework. STEV includes a new Flat Scheme to tackle the inconsistent data shape issue, which extends the graph-based spatio-temporal modeling architecture into 1D space by flattening the 2D samples along the variable dimension, making the model variable-scale-agnostic while still preserving dynamic spatial correlations through a holistic graph. We introduce a novel Spatio-temporal Focal Learning strategy that incorporates a negative filter to resolve potential conflicts between contrastive learning and graph representation, and a focal contrastive loss as its core to guide the framework to focus on optimizing the expanding variables. We benchmark EVTSF performance using three real-world datasets and compare it against three potential solutions employing SOTA MTSF models tailored for EVSTF. Experimental results show that STEV significantly outperforms its competitors, particularly on expanding variables. Notably, STEV, with only 5% of observations from the expanding period, is on par with SOTA MTSF models trained with complete observations. Further exploration of various expanding strategies underscores the generalizability of STEV in real-world applications.
- [523] arXiv:2502.15455 (replaced) [pdf, html, other]
-
Title: R-LoRA: Randomized Multi-Head LoRA for Efficient Multi-Task LearningComments: 8 pages, 7 figuresSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Fine-tuning large language models (LLMs) is computationally expensive, and Low-Rank Adaptation (LoRA) provides a cost-effective solution by approximating weight updates through low-rank matrices. In real-world scenarios, LLMs are fine-tuned on data from multiple domains to perform tasks across various fields, embodying multi-task learning (MTL). LoRA often underperforms in such complex scenarios. To enhance LoRA's capability in multi-task learning, we propose R-LoRA, which incorporates Multi-Head Randomization. Multi-Head Randomization diversifies the head matrices through Multi-Head Dropout and Multi-Head Random Initialization, enabling more efficient learning of task-specific features while maintaining shared knowledge representation. Our approach not only improves performance in MTL but also reduces GPU memory usage and training time. Experiments show that R-LoRA's gains stem from increased diversity in the head matrices, demonstrating its effectiveness for multi-task learning. The code is available at this https URL
- [524] arXiv:2502.15798 (replaced) [pdf, html, other]
-
Title: MaxSup: Overcoming Representation Collapse in Label SmoothingComments: 24 pages, 15 tables, 5 figures. Preliminary work under review. Do not distributeSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
Label Smoothing (LS) is widely adopted to reduce overconfidence in neural network predictions and improve generalization. Despite these benefits, recent studies reveal two critical issues with LS. First, LS induces overconfidence in misclassified samples. Second, it compacts feature representations into overly tight clusters, diluting intra-class diversity, although the precise cause of this phenomenon remained elusive. In this paper, we analytically decompose the LS-induced loss, exposing two key terms: (i) a regularization term that dampens overconfidence only when the prediction is correct, and (ii) an error-amplification term that arises under misclassifications. This latter term compels the network to reinforce incorrect predictions with undue certainty, exacerbating representation collapse. To address these shortcomings, we propose Max Suppression (MaxSup), which applies uniform regularization to both correct and incorrect predictions by penalizing the top-1 logit rather than the ground-truth logit. Through extensive feature-space analyses, we show that MaxSup restores intra-class variation and sharpens inter-class boundaries. Experiments on large-scale image classification and multiple downstream tasks confirm that MaxSup is a more robust alternative to LS, consistently reducing overconfidence while preserving richer feature representations. Code is available at: this https URL
- [525] arXiv:2502.17019 (replaced) [pdf, html, other]
-
Title: Erwin: A Tree-based Hierarchical Transformer for Large-scale Physical SystemsComments: Accepted to ICML 2025. Code: this https URLSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
Large-scale physical systems defined on irregular grids pose significant scalability challenges for deep learning methods, especially in the presence of long-range interactions and multi-scale coupling. Traditional approaches that compute all pairwise interactions, such as attention, become computationally prohibitive as they scale quadratically with the number of nodes. We present Erwin, a hierarchical transformer inspired by methods from computational many-body physics, which combines the efficiency of tree-based algorithms with the expressivity of attention mechanisms. Erwin employs ball tree partitioning to organize computation, which enables linear-time attention by processing nodes in parallel within local neighborhoods of fixed size. Through progressive coarsening and refinement of the ball tree structure, complemented by a novel cross-ball interaction mechanism, it captures both fine-grained local details and global features. We demonstrate Erwin's effectiveness across multiple domains, including cosmology, molecular dynamics, PDE solving, and particle fluid dynamics, where it consistently outperforms baseline methods both in accuracy and computational efficiency.
- [526] arXiv:2502.18137 (replaced) [pdf, html, other]
-
Title: SpargeAttention: Accurate and Training-free Sparse Attention Accelerating Any Model InferenceComments: @inproceedings{zhang2025spargeattn, title={Spargeattn: Accurate sparse attention accelerating any model inference}, author={Zhang, Jintao and Xiang, Chendong and Huang, Haofeng and Wei, Jia and Xi, Haocheng and Zhu, Jun and Chen, Jianfei}, booktitle={International Conference on Machine Learning (ICML)}, year={2025} }Journal-ref: Proceedings of the 42 nd International Conference on Machine Learning, PMLR 267, 2025 (ICML 2025)Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Performance (cs.PF)
An efficient attention implementation is essential for large models due to its quadratic time complexity. Fortunately, attention commonly exhibits sparsity, i.e., many values in the attention map are near zero, allowing for the omission of corresponding computations. Many studies have utilized the sparse pattern to accelerate attention. However, most existing works focus on optimizing attention within specific models by exploiting certain sparse patterns of the attention map. A universal sparse attention that guarantees both the speedup and end-to-end performance of diverse models remains elusive. In this paper, we propose SpargeAttn, a universal sparse and quantized attention for any model. Our method uses a two-stage online filter: in the first stage, we rapidly and accurately predict the attention map, enabling the skip of some matrix multiplications in attention. In the second stage, we design an online softmax-aware filter that incurs no extra overhead and further skips some matrix multiplications. Experiments show that our method significantly accelerates diverse models, including language, image, and video generation, without sacrificing end-to-end metrics. The codes are available at this https URL.
- [527] arXiv:2502.19726 (replaced) [pdf, html, other]
-
Title: Tokens for Learning, Tokens for Unlearning: Mitigating Membership Inference Attacks in Large Language Models via Dual-Purpose TrainingComments: ACL'25 (Findings)Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Large language models (LLMs) have become the backbone of modern natural language processing but pose privacy concerns about leaking sensitive training data. Membership inference attacks (MIAs), which aim to infer whether a sample is included in a model's training dataset, can serve as a foundation for broader privacy threats. Existing defenses designed for traditional classification models do not account for the sequential nature of text data. As a result, they either require significant computational resources or fail to effectively mitigate privacy risks in LLMs. In this work, we propose \methodname, a lightweight yet effective empirical privacy defense for protecting training data of language models by leveraging token-specific characteristics. By analyzing token dynamics during training, we propose a token selection strategy that categorizes tokens into hard tokens for learning and memorized tokens for unlearning. Subsequently, our training-phase defense optimizes a novel dual-purpose token-level loss to achieve a Pareto-optimal balance between utility and privacy. Extensive experiments demonstrate that our approach not only provides strong protection against MIAs but also improves language modeling performance by around 10\% across various LLM architectures and datasets compared to the baselines.
- [528] arXiv:2502.20317 (replaced) [pdf, html, other]
-
Title: Mixture of Structural-and-Textual Retrieval over Text-rich Graph Knowledge BasesYongjia Lei, Haoyu Han, Ryan A. Rossi, Franck Dernoncourt, Nedim Lipka, Mahantesh M Halappanavar, Jiliang Tang, Yu WangSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
Text-rich Graph Knowledge Bases (TG-KBs) have become increasingly crucial for answering queries by providing textual and structural knowledge. However, current retrieval methods often retrieve these two types of knowledge in isolation without considering their mutual reinforcement and some hybrid methods even bypass structural retrieval entirely after neighboring aggregation. To fill in this gap, we propose a Mixture of Structural-and-Textual Retrieval (MoR) to retrieve these two types of knowledge via a Planning-Reasoning-Organizing framework. In the Planning stage, MoR generates textual planning graphs delineating the logic for answering queries. Following planning graphs, in the Reasoning stage, MoR interweaves structural traversal and textual matching to obtain candidates from TG-KBs. In the Organizing stage, MoR further reranks fetched candidates based on their structural trajectory. Extensive experiments demonstrate the superiority of MoR in harmonizing structural and textual retrieval with insights, including uneven retrieving performance across different query logics and the benefits of integrating structural trajectories for candidate reranking. Our code is available at this https URL.
- [529] arXiv:2503.00206 (replaced) [pdf, html, other]
-
Title: Quantifying First-Order Markov Violations in Noisy Reinforcement Learning: A Causal Discovery ApproachComments: Under review for Neural Information Processing Systems 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Reinforcement learning (RL) methods frequently assume that each new observation completely reflects the environment's state, thereby guaranteeing Markovian (one-step) transitions. In practice, partial observability or sensor/actuator noise often invalidates this assumption. This paper proposes a systematic methodology for detecting such violations, combining a partial correlation-based causal discovery process (PCMCI) with a novel Markov Violation score (MVS). The MVS measures multi-step dependencies that emerge when noise or incomplete state information disrupts the Markov property.
Classic control tasks (CartPole, Pendulum, Acrobot) serve as examples to illustrate how targeted noise and dimension omissions affect both RL performance and measured Markov consistency. Surprisingly, even substantial observation noise sometimes fails to induce strong multi-lag dependencies in certain domains (e.g., Acrobot). In contrast, dimension-dropping investigations show that excluding some state variables (e.g., angular velocities in CartPole and Pendulum) significantly reduces returns and increases MVS, while removing other dimensions has minimal impact.
These findings emphasize the importance of locating and safeguarding the most causally essential dimensions in order to preserve effective single-step learning. By integrating partial correlation tests with RL performance outcomes, the proposed approach precisely identifies when and where the Markov assumption is violated. This framework offers a principled mechanism for developing robust policies, informing representation learning, and addressing partial observability in real-world RL scenarios. All code and experimental logs are accessible for reproducibility (this https URL). - [530] arXiv:2503.01450 (replaced) [pdf, html, other]
-
Title: POPGym Arcade: Parallel Pixelated POMDPsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Robotics (cs.RO)
We present the POPGym Arcade, a collection of hardware-accelerated, pixel-based environments with shared observation and action spaces. Each environment includes fully and partially observable variants, enabling counterfactual studies on partial observability. We also introduce mathematical tools for analyzing policies under partial observability, which reveal how agents recall past information to make decisions. Our analysis shows (1) that controlling for partial observability is critical and (2) that agents with long-term memory learn brittle policies that struggle to generalize. Finally, we demonstrate that recurrent policies can be "poisoned" by old, out-of-distribution observations, with implications for sim-to-real transfer, imitation learning, and offline reinforcement learning.
- [531] arXiv:2503.01461 (replaced) [pdf, html, other]
-
Title: Marco-o1 v2: Towards Widening The Distillation Bottleneck for Reasoning ModelsHuifeng Yin, Yu Zhao, Minghao Wu, Xuanfan Ni, Bo Zeng, Hao Wang, Tianqi Shi, Liangying Shao, Chenyang Lyu, Longyue Wang, Weihua Luo, Kaifu ZhangSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Large Reasoning Models(LRMs) such as OpenAI o1 and DeepSeek-R1 have shown remarkable reasoning capabilities by scaling test-time compute and generating long Chain-of-Thought(CoT). Distillation--post-training on LRMs-generated data--is a straightforward yet effective method to enhance the reasoning abilities of smaller models, but faces a critical bottleneck: we found that distilled long CoT data poses learning difficulty for small models and leads to the inheritance of biases (i.e. over-thinking) when using Supervised Fine-tuning (SFT) and Reinforcement Learning (RL) methods. To alleviate this bottleneck, we propose constructing tree-based CoT data from scratch via Monte Carlo Tree Search(MCTS). We then exploit a set of CoT-aware approaches, including Thoughts Length Balance, Fine-grained DPO, and Joint Post-training Objective, to enhance SFT and RL on the constructed data. We conduct evaluation on various benchmarks such as math (GSM8K, MATH, AIME). instruction-following (Multi-IF) and planning (Blocksworld), results demonstrate our approaches substantially improve the reasoning performance of distilled models compared to standard distilled models via reducing the hallucinations in long-time thinking. The project homepage is this https URL.
- [532] arXiv:2503.01891 (replaced) [pdf, html, other]
-
Title: MMSciBench: Benchmarking Language Models on Chinese Multimodal Scientific ProblemsComments: Accepted to the Findings of the Association for Computational Linguistics (ACL 2025)Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Recent advances in large language models (LLMs) and vision-language models (LVLMs) have shown promise across many tasks, yet their scientific reasoning capabilities remain untested, particularly in multimodal settings. We present MMSciBench, a benchmark for evaluating mathematical and physical reasoning through text-only and text-image formats, with human-annotated difficulty levels, solutions with detailed explanations, and taxonomic mappings. Evaluation of state-of-the-art models reveals significant limitations, with even the best model achieving only \textbf{63.77\%} accuracy and particularly struggling with visual reasoning tasks. Our analysis exposes critical gaps in complex reasoning and visual-textual integration, establishing MMSciBench as a rigorous standard for measuring progress in multimodal scientific understanding. The code for MMSciBench is open-sourced at GitHub, and the dataset is available at Hugging Face.
- [533] arXiv:2503.03043 (replaced) [pdf, html, other]
-
Title: Leveraging Randomness in Model and Data Partitioning for Privacy AmplificationSubjects: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
We study how inherent randomness in the training process -- where each sample (or client in federated learning) contributes only to a randomly selected portion of training -- can be leveraged for privacy amplification. This includes (1) data partitioning, where a sample participates in only a subset of training iterations, and (2) model partitioning, where a sample updates only a subset of the model parameters. We apply our framework to model parallelism in federated learning, where each client updates a randomly selected subnetwork to reduce memory and computational overhead, and show that existing methods, e.g. model splitting or dropout, provide a significant privacy amplification gain not captured by previous privacy analysis techniques. Additionally, we introduce Balanced Iteration Subsampling, a new data partitioning method where each sample (or client) participates in a fixed number of training iterations. We show that this method yields stronger privacy amplification than Poisson (i.i.d.) sampling of data (or clients). Our results demonstrate that randomness in the training process, which is structured rather than i.i.d. and interacts with data in complex ways, can be systematically leveraged for significant privacy amplification.
- [534] arXiv:2503.04363 (replaced) [pdf, html, other]
-
Title: Causally Reliable Concept Bottleneck ModelsGiovanni De Felice, Arianna Casanova Flores, Francesco De Santis, Silvia Santini, Johannes Schneider, Pietro Barbiero, Alberto TermineSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Concept-based models are an emerging paradigm in deep learning that constrains the inference process to operate through human-interpretable variables, facilitating explainability and human interaction. However, these architectures, on par with popular opaque neural models, fail to account for the true causal mechanisms underlying the target phenomena represented in the data. This hampers their ability to support causal reasoning tasks, limits out-of-distribution generalization, and hinders the implementation of fairness constraints. To overcome these issues, we propose Causally reliable Concept Bottleneck Models (C$^2$BMs), a class of concept-based architectures that enforce reasoning through a bottleneck of concepts structured according to a model of the real-world causal mechanisms. We also introduce a pipeline to automatically learn this structure from observational data and unstructured background knowledge (e.g., scientific literature). Experimental evidence suggests that C$^2$BMs are more interpretable, causally reliable, and improve responsiveness to interventions w.r.t. standard opaque and concept-based models, while maintaining their accuracy.
- [535] arXiv:2503.04992 (replaced) [pdf, html, other]
-
Title: Wanda++: Pruning Large Language Models via Regional GradientsYifan Yang, Kai Zhen, Bhavana Ganesh, Aram Galstyan, Goeric Huybrechts, Markus Müller, Jonas M. Kübler, Rupak Vignesh Swaminathan, Athanasios Mouchtaris, Sravan Babu Bodapati, Nathan Susanj, Zheng Zhang, Jack FitzGerald, Abhishek KumarComments: Paper accepted at ACL 2025 FindingsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Large Language Models (LLMs) pruning seeks to remove unimportant weights for inference speedup with minimal accuracy impact. However, existing methods often suffer from accuracy degradation without full-model sparsity-aware fine-tuning. This paper presents Wanda++, a novel pruning framework that outperforms the state-of-the-art methods by utilizing decoder-block-level \textbf{regional} gradients. Specifically, Wanda++ improves the pruning score with regional gradients for the first time and proposes an efficient regional optimization method to minimize pruning-induced output discrepancies between the dense and sparse decoder output. Notably, Wanda++ improves perplexity by up to 32\% over Wanda in the language modeling task and generalizes effectively to downstream tasks. Moreover, despite updating weights with regional optimization, Wanda++ remains orthogonal to sparsity-aware fine-tuning, further reducing perplexity with LoRA in great extend. Our approach is lightweight, pruning a 7B LLaMA model in under 10 minutes on a single H100 GPU.
- [536] arXiv:2503.05315 (replaced) [pdf, html, other]
-
Title: LoRACode: LoRA Adapters for Code EmbeddingsComments: Accepted at the Deep Learning for Code (DL4C) Workshop at ICLR 2025Subjects: Machine Learning (cs.LG); Information Retrieval (cs.IR); Software Engineering (cs.SE)
Code embeddings are essential for semantic code search; however, current approaches often struggle to capture the precise syntactic and contextual nuances inherent in code. Open-source models such as CodeBERT and UniXcoder exhibit limitations in scalability and efficiency, while high-performing proprietary systems impose substantial computational costs. We introduce a parameter-efficient fine-tuning method based on Low-Rank Adaptation (LoRA) to construct task-specific adapters for code retrieval. Our approach reduces the number of trainable parameters to less than two percent of the base model, enabling rapid fine-tuning on extensive code corpora (2 million samples in 25 minutes on two H100 GPUs). Experiments demonstrate an increase of up to 9.1% in Mean Reciprocal Rank (MRR) for Code2Code search, and up to 86.69% for Text2Code search tasks across multiple programming languages. Distinction in task-wise and language-wise adaptation helps explore the sensitivity of code retrieval for syntactical and linguistic variations. To foster research in this area, we make our code and pre-trained models publicly available.
- [537] arXiv:2503.07580 (replaced) [pdf, html, other]
-
Title: BOPO: Neural Combinatorial Optimization via Best-anchored and Objective-guided Preference OptimizationComments: This paper has been accepted by ICML 2025Subjects: Machine Learning (cs.LG)
Neural Combinatorial Optimization (NCO) has emerged as a promising approach for NP-hard problems. However, prevailing RL-based methods suffer from low sample efficiency due to sparse rewards and underused solutions. We propose Best-anchored and Objective-guided Preference Optimization (BOPO), a training paradigm that leverages solution preferences via objective values. It introduces: (1) a best-anchored preference pair construction for better explore and exploit solutions, and (2) an objective-guided pairwise loss function that adaptively scales gradients via objective differences, removing reliance on reward models or reference policies. Experiments on Job-shop Scheduling Problem (JSP), Traveling Salesman Problem (TSP), and Flexible Job-shop Scheduling Problem (FJSP) show BOPO outperforms state-of-the-art neural methods, reducing optimality gaps impressively with efficient inference. BOPO is architecture-agnostic, enabling seamless integration with existing NCO models, and establishes preference optimization as a principled framework for combinatorial optimization.
- [538] arXiv:2503.09532 (replaced) [pdf, html, other]
-
Title: SAEBench: A Comprehensive Benchmark for Sparse Autoencoders in Language Model InterpretabilityAdam Karvonen, Can Rager, Johnny Lin, Curt Tigges, Joseph Bloom, David Chanin, Yeu-Tong Lau, Eoin Farrell, Callum McDougall, Kola Ayonrinde, Demian Till, Matthew Wearden, Arthur Conmy, Samuel Marks, Neel NandaComments: Accepted to ICML 2025 main conferenceSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Sparse autoencoders (SAEs) are a popular technique for interpreting language model activations, and there is extensive recent work on improving SAE effectiveness. However, most prior work evaluates progress using unsupervised proxy metrics with unclear practical relevance. We introduce SAEBench, a comprehensive evaluation suite that measures SAE performance across eight diverse metrics, spanning interpretability, feature disentanglement and practical applications like unlearning. To enable systematic comparison, we open-source a suite of over 200 SAEs across eight recently proposed SAE architectures and training algorithms. Our evaluation reveals that gains on proxy metrics do not reliably translate to better practical performance. For instance, while Matryoshka SAEs slightly underperform on existing proxy metrics, they substantially outperform other architectures on feature disentanglement metrics; moreover, this advantage grows with SAE scale. By providing a standardized framework for measuring progress in SAE development, SAEBench enables researchers to study scaling trends and make nuanced comparisons between different SAE architectures and training methodologies. Our interactive interface enables researchers to flexibly visualize relationships between metrics across hundreds of open-source SAEs at: this http URL
- [539] arXiv:2503.13509 (replaced) [pdf, html, other]
-
Title: MentalChat16K: A Benchmark Dataset for Conversational Mental Health AssistanceJia Xu, Tianyi Wei, Bojian Hou, Patryk Orzechowski, Shu Yang, Ruochen Jin, Rachael Paulbeck, Joost Wagenaar, George Demiris, Li ShenSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
We introduce MentalChat16K, an English benchmark dataset combining a synthetic mental health counseling dataset and a dataset of anonymized transcripts from interventions between Behavioral Health Coaches and Caregivers of patients in palliative or hospice care. Covering a diverse range of conditions like depression, anxiety, and grief, this curated dataset is designed to facilitate the development and evaluation of large language models for conversational mental health assistance. By providing a high-quality resource tailored to this critical domain, MentalChat16K aims to advance research on empathetic, personalized AI solutions to improve access to mental health support services. The dataset prioritizes patient privacy, ethical considerations, and responsible data usage. MentalChat16K presents a valuable opportunity for the research community to innovate AI technologies that can positively impact mental well-being. The dataset is available at this https URL and the code and documentation are hosted on GitHub at this https URL.
- [540] arXiv:2503.16072 (replaced) [pdf, html, other]
-
Title: Redefining Toxicity: An Objective and Context-Aware Approach for Stress-Level-Based DetectionSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Most toxicity detection models treat toxicity as an intrinsic property of text, overlooking the role of context in shaping its impact. Drawing on interdisciplinary research, we reconceptualise toxicity as a socially emergent stress signal. We introduce a new framework for toxicity detection, including a formal definition and metric, and validate our approach on a novel dataset, demonstrating improved contextual sensitivity and adaptability.
- [541] arXiv:2504.00010 (replaced) [pdf, html, other]
-
Title: LayerCraft: Enhancing Text-to-Image Generation with CoT Reasoning and Layered Object IntegrationComments: 26 pagesSubjects: Machine Learning (cs.LG); Graphics (cs.GR); Multiagent Systems (cs.MA)
Text-to-image (T2I) generation has made remarkable progress, yet existing systems still lack intuitive control over spatial composition, object consistency, and multi-step editing. We present $\textbf{LayerCraft}$, a modular framework that uses large language models (LLMs) as autonomous agents to orchestrate structured, layered image generation and editing. LayerCraft supports two key capabilities: (1) $\textit{structured generation}$ from simple prompts via chain-of-thought (CoT) reasoning, enabling it to decompose scenes, reason about object placement, and guide composition in a controllable, interpretable manner; and (2) $\textit{layered object integration}$, allowing users to insert and customize objects -- such as characters or props -- across diverse images or scenes while preserving identity, context, and style. The system comprises a coordinator agent, the $\textbf{ChainArchitect}$ for CoT-driven layout planning, and the $\textbf{Object Integration Network (OIN)}$ for seamless image editing using off-the-shelf T2I models without retraining. Through applications like batch collage editing and narrative scene generation, LayerCraft empowers non-experts to iteratively design, customize, and refine visual content with minimal manual effort. Code will be released at this https URL.
- [542] arXiv:2504.03801 (replaced) [pdf, html, other]
-
Title: Semantic-guided Representation Learning for Multi-Label RecognitionComments: Accepted in ICME2025 Oral (15% of all submissions)Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Multi-label Recognition (MLR) involves assigning multiple labels to each data instance in an image, offering advantages over single-label classification in complex scenarios. However, it faces the challenge of annotating all relevant categories, often leading to uncertain annotations, such as unseen or incomplete labels. Recent Vision and Language Pre-training (VLP) based methods have made significant progress in tackling zero-shot MLR tasks by leveraging rich vision-language correlations. However, the correlation between multi-label semantics has not been fully explored, and the learned visual features often lack essential semantic information. To overcome these limitations, we introduce a Semantic-guided Representation Learning approach (SigRL) that enables the model to learn effective visual and textual representations, thereby improving the downstream alignment of visual images and categories. Specifically, we first introduce a graph-based multi-label correlation module (GMC) to facilitate information exchange between labels, enriching the semantic representation across the multi-label texts. Next, we propose a Semantic Visual Feature Reconstruction module (SVFR) to enhance the semantic information in the visual representation by integrating the learned textual representation during reconstruction. Finally, we optimize the image-text matching capability of the VLP model using both local and global features to achieve zero-shot MLR. Comprehensive experiments are conducted on several MLR benchmarks, encompassing both zero-shot MLR (with unseen labels) and single positive multi-label learning (with limited labels), demonstrating the superior performance of our approach compared to state-of-the-art methods. The code is available at this https URL.
- [543] arXiv:2504.04799 (replaced) [pdf, html, other]
-
Title: Topological Schrödinger Bridge MatchingComments: ICLR 2025 Spotlight, 42 pagesSubjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Given two boundary distributions, the Schrödinger Bridge (SB) problem seeks the ``most likely`` random evolution between them with respect to a reference process. It has revealed rich connections to recent machine learning methods for generative modeling and distribution matching. While these methods perform well in Euclidean domains, they are not directly applicable to topological domains such as graphs and simplicial complexes, which are crucial for data defined over network entities, such as node signals and edge flows. In this work, we propose the Topological Schrödinger Bridge problem (TSBP) for matching signal distributions on a topological domain. We set the reference process to follow some linear tractable topology-aware stochastic dynamics such as topological heat diffusion. For the case of Gaussian boundary distributions, we derive a closed-form topological SB (TSB) in terms of its time-marginal and stochastic differential. In the general case, leveraging the well-known result, we show that the optimal process follows the forward-backward topological dynamics governed by some unknowns. Building on these results, we develop TSB-based models for matching topological signals by parameterizing the unknowns in the optimal process as (topological) neural networks and learning them through likelihood training. We validate the theoretical results and demonstrate the practical applications of TSB-based models on both synthetic and real-world networks, emphasizing the role of topology. Additionally, we discuss the connections of TSB-based models to other emerging models, and outline future directions for topological signal matching.
- [544] arXiv:2504.13075 (replaced) [pdf, html, other]
-
Title: An All-Atom Generative Model for Designing Protein ComplexesComments: ICML 2025 camera-readySubjects: Machine Learning (cs.LG)
Proteins typically exist in complexes, interacting with other proteins or biomolecules to perform their specific biological roles. Research on single-chain protein modeling has been extensively and deeply explored, with advancements seen in models like the series of ESM and AlphaFold2. Despite these developments, the study and modeling of multi-chain proteins remain largely uncharted, though they are vital for understanding biological functions. Recognizing the importance of these interactions, we introduce APM (All-Atom Protein Generative Model), a model specifically designed for modeling multi-chain proteins. By integrating atom-level information and leveraging data on multi-chain proteins, APM is capable of precisely modeling inter-chain interactions and designing protein complexes with binding capabilities from scratch. It also performs folding and inverse-folding tasks for multi-chain proteins. Moreover, APM demonstrates versatility in downstream applications: it achieves enhanced performance through supervised fine-tuning (SFT) while also supporting zero-shot sampling in certain tasks, achieving state-of-the-art results. We released our code at this https URL.
- [545] arXiv:2504.13101 (replaced) [pdf, other]
-
Title: Position: An Empirically Grounded Identifiability Theory Will Accelerate Self-Supervised Learning ResearchComments: ICML2025 camera readySubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Self-Supervised Learning (SSL) powers many current AI systems. As research interest and investment grow, the SSL design space continues to expand. The Platonic view of SSL, following the Platonic Representation Hypothesis (PRH), suggests that despite different methods and engineering approaches, all representations converge to the same Platonic ideal. However, this phenomenon lacks precise theoretical explanation. By synthesizing evidence from Identifiability Theory (IT), we show that the PRH can emerge in SSL. However, current IT cannot explain SSL's empirical success. To bridge the gap between theory and practice, we propose expanding IT into what we term Singular Identifiability Theory (SITh), a broader theoretical framework encompassing the entire SSL pipeline. SITh would allow deeper insights into the implicit data assumptions in SSL and advance the field towards learning more interpretable and generalizable representations. We highlight three critical directions for future research: 1) training dynamics and convergence properties of SSL; 2) the impact of finite samples, batch size, and data diversity; and 3) the role of inductive biases in architecture, augmentations, initialization schemes, and optimizers.
- [546] arXiv:2504.17004 (replaced) [pdf, html, other]
-
Title: (Im)possibility of Automated Hallucination Detection in Large Language ModelsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (stat.ML)
Is automated hallucination detection possible? In this work, we introduce a theoretical framework to analyze the feasibility of automatically detecting hallucinations produced by large language models (LLMs). Inspired by the classical Gold-Angluin framework for language identification and its recent adaptation to language generation by Kleinberg and Mullainathan, we investigate whether an algorithm, trained on examples drawn from an unknown target language $K$ (selected from a countable collection) and given access to an LLM, can reliably determine whether the LLM's outputs are correct or constitute hallucinations.
First, we establish an equivalence between hallucination detection and the classical task of language identification. We prove that any hallucination detection method can be converted into a language identification method, and conversely, algorithms solving language identification can be adapted for hallucination detection. Given the inherent difficulty of language identification, this implies that hallucination detection is fundamentally impossible for most language collections if the detector is trained using only correct examples from the target language.
Second, we show that the use of expert-labeled feedback, i.e., training the detector with both positive examples (correct statements) and negative examples (explicitly labeled incorrect statements), dramatically changes this conclusion. Under this enriched training regime, automated hallucination detection becomes possible for all countable language collections.
These results highlight the essential role of expert-labeled examples in training hallucination detectors and provide theoretical support for feedback-based methods, such as reinforcement learning with human feedback (RLHF), which have proven critical for reliable LLM deployment. - [547] arXiv:2504.19583 (replaced) [pdf, other]
-
Title: Graph-Based Spectral Decomposition for Parameter Coordination in Language Model Fine-TuningSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
This paper proposes a parameter collaborative optimization algorithm for large language models, enhanced with graph spectral analysis. The goal is to improve both fine-tuning efficiency and structural awareness during training. In the proposed method, the parameters of a pre-trained language model are treated as nodes in a graph. A weighted graph is constructed, and Laplacian spectral decomposition is applied to enable frequency-domain modeling and structural representation of the parameter space. Based on this structure, a joint loss function is designed. It combines the task loss with a spectral regularization term to facilitate collaborative updates among parameters. In addition, a spectral filtering mechanism is introduced during the optimization phase. This mechanism adjusts gradients in a structure-aware manner, enhancing the model's training stability and convergence behavior. The method is evaluated on multiple tasks, including traditional fine-tuning comparisons, few-shot generalization tests, and convergence speed analysis. In all settings, the proposed approach demonstrates superior performance. The experimental results confirm that the spectral collaborative optimization framework effectively reduces parameter perturbations and improves fine-tuning quality while preserving overall model performance. This work contributes significantly to the field of artificial intelligence by advancing parameter-efficient training methodologies for large-scale models, reinforcing the importance of structural signal processing in deep learning optimization, and offering a robust, generalizable framework for enhancing language model adaptability and performance.
- [548] arXiv:2505.00812 (replaced) [pdf, html, other]
-
Title: Handling Label Noise via Instance-Level Difficulty Modeling and Dynamic OptimizationSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Recent studies indicate that deep neural networks degrade in generalization performance under noisy supervision. Existing methods focus on isolating clean subsets or correcting noisy labels, facing limitations such as high computational costs, heavy hyperparameter tuning process, and coarse-grained optimization. To address these challenges, we propose a novel two-stage noisy learning framework that enables instance-level optimization through a dynamically weighted loss function, avoiding hyperparameter tuning. To obtain stable and accurate information about noise modeling, we introduce a simple yet effective metric, termed wrong event, which dynamically models the cleanliness and difficulty of individual samples while maintaining computational costs. Our framework first collects wrong event information and builds a strong base model. Then we perform noise-robust training on the base model, using a probabilistic model to handle the wrong event information of samples. Experiments on five synthetic and real-world LNL benchmarks demonstrate our method surpasses state-of-the-art methods in performance, achieves a nearly 75% reduction in computational time and improves model scalability.
- [549] arXiv:2505.01199 (replaced) [pdf, html, other]
-
Title: CaReAQA: A Cardiac and Respiratory Audio Question Answering Model for Open-Ended Diagnostic ReasoningComments: Accepted at AHLI CHIL 2025Subjects: Machine Learning (cs.LG)
Medical audio signals, such as heart and lung sounds, play a crucial role in clinical diagnosis. However, analyzing these signals remains challenging: traditional methods rely on handcrafted features or supervised deep learning models that demand extensive labeled datasets, limiting their scalability and applicability. To address these issues, we propose CaReAQA, an audio-language model that integrates a foundation audio model with the reasoning capabilities of large language models, enabling clinically relevant, open-ended diagnostic responses. Alongside CaReAQA, we introduce CaReSound, a benchmark dataset of annotated medical audio recordings enriched with metadata and paired question-answer examples, intended to drive progress in diagnostic reasoning research. Evaluation results show that CaReAQA achieves 86.2% accuracy on open-ended diagnostic reasoning tasks, outperforming baseline models. It also generalizes well to closed-ended classification tasks, achieving an average accuracy of 56.9% on unseen datasets. Our findings show how audio-language integration and reasoning advances medical diagnostics, enabling efficient AI systems for clinical decision support.
- [550] arXiv:2505.01892 (replaced) [pdf, html, other]
-
Title: OODTE: A Differential Testing Engine for the ONNX OptimizerComments: 12 pages, 3 figures, 3 tablesSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Software Engineering (cs.SE); Systems and Control (eess.SY)
With over 700 stars on GitHub and being part of the official ONNX repository, the ONNX Optimizer is the default tool for applying graph-based optimizations to ONNX models. Despite its widespread use, its ability to maintain model accuracy during optimization has not been thoroughly investigated. In this work, we present OODTE, a utility designed to automatically and comprehensively evaluate the correctness of the ONNX Optimizer. OODTE adopts a straightforward yet powerful differential testing and evaluation methodology, which can be readily adapted for use with other compiler optimizers. Specifically, OODTE takes a collection of ONNX models, applies optimizations, and executes both the original and optimized versions across a user-defined input set, automatically capturing any issues encountered during optimization. When discrepancies in accuracy arise, OODTE iteratively isolates the responsible optimization pass by repeating the process at a finer granularity. We applied OODTE to 130 well-known models from the official ONNX Model Hub, spanning diverse tasks including classification, object detection, semantic segmentation, text summarization, question answering, and sentiment analysis. Our evaluation revealed that 9.2% of the model instances either caused the optimizer to crash or led to the generation of invalid models using default optimization strategies. Additionally, 30% of classification models and 16.6% of object detection and segmentation models exhibited differing outputs across original and optimized versions, whereas models focused on text-related tasks were generally robust to optimization. OODTE uncovered 15 issues-14 previously unknown-affecting 9 of 47 optimization passes and the optimizer overall. All issues were reported to the ONNX Optimizer team. OODTE offers a simple but effective framework for validating AI model optimizers, applicable beyond the ONNX ecosystem.
- [551] arXiv:2505.02515 (replaced) [pdf, html, other]
-
Title: FedSDAF: Leveraging Source Domain Awareness for Enhanced Federated Domain GeneralizationSubjects: Machine Learning (cs.LG)
Traditional domain generalization approaches predominantly focus on leveraging target domain-aware features while overlooking the critical role of source domain-specific characteristics, particularly in federated settings with inherent data isolation. To address this gap, we propose the Federated Source Domain Awareness Framework (FedSDAF), the first method to systematically exploit source domain-aware features for enhanced federated domain generalization (FedDG). The FedSDAF framework consists of two synergistic components: the Domain-Invariant Adapter, which preserves critical domain-invariant features, and the Domain-Aware Adapter, which extracts and integrates source domain-specific knowledge using a Multihead Self-Attention mechanism (MHSA). Furthermore, we introduce a bidirectional knowledge distillation mechanism that fosters knowledge sharing among clients while safeguarding privacy. Our approach represents the first systematic exploitation of source domain-aware features, resulting in significant advancements in model generalization this http URL experiments on four standard benchmarks (OfficeHome, PACS, VLCS, and DomainNet) show that our method consistently surpasses state-of-the-art federated domain generalization approaches, with accuracy gains of 5.2-13.8%. The source code is available at this https URL.
- [552] arXiv:2505.03793 (replaced) [pdf, html, other]
-
Title: LENSLLM: Unveiling Fine-Tuning Dynamics for LLM SelectionComments: Accepted by ICML'2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
The proliferation of open-sourced Large Language Models (LLMs) and diverse downstream tasks necessitates efficient model selection, given the impracticality of fine-tuning all candidates due to computational constraints. Despite the recent advances in LLM selection, a fundamental research question largely remains nascent: how can we model the dynamic behaviors of LLMs during fine-tuning, thereby enhancing our understanding of their generalization performance across diverse downstream tasks? In this work, we propose a novel theoretical framework that provides a proper lens to assess the generalization capabilities of LLMs, thereby enabling accurate and efficient LLM selection for downstream applications. In particular, we first derive a PAC-Bayesian Generalization Bound that unveils fine-tuning dynamics of LLMs and then introduce LENSLLM, a Neural Tangent Kernel (NTK)-based Rectified Scaling Model that enables accurate performance predictions across diverse tasks while maintaining computational efficiency. Extensive empirical results on 3 large-scale benchmarks demonstrate that our model achieves up to 91.1% accuracy and reduces up to 88.5% computational cost in LLM selection, outperforming 5 state-of-the-art methods. We open-source our proposed LENSLLM model and corresponding results at this http URL.
- [553] arXiv:2505.04608 (replaced) [pdf, html, other]
-
Title: WATCH: Adaptive Monitoring for AI Deployments via Weighted-Conformal MartingalesComments: To be published in The International Conference on Machine Learning (ICML), 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Responsibly deploying artificial intelligence (AI) / machine learning (ML) systems in high-stakes settings arguably requires not only proof of system reliability, but also continual, post-deployment monitoring to quickly detect and address any unsafe behavior. Methods for nonparametric sequential testing -- especially conformal test martingales (CTMs) and anytime-valid inference -- offer promising tools for this monitoring task. However, existing approaches are restricted to monitoring limited hypothesis classes or ``alarm criteria'' (e.g., detecting data shifts that violate certain exchangeability or IID assumptions), do not allow for online adaptation in response to shifts, and/or cannot diagnose the cause of degradation or alarm. In this paper, we address these limitations by proposing a weighted generalization of conformal test martingales (WCTMs), which lay a theoretical foundation for online monitoring for any unexpected changepoints in the data distribution while controlling false-alarms. For practical applications, we propose specific WCTM algorithms that adapt online to mild covariate shifts (in the marginal input distribution), quickly detect harmful shifts, and diagnose those harmful shifts as concept shifts (in the conditional label distribution) or extreme (out-of-support) covariate shifts that cannot be easily adapted to. On real-world datasets, we demonstrate improved performance relative to state-of-the-art baselines.
- [554] arXiv:2505.05702 (replaced) [pdf, html, other]
-
Title: Hypergraph Neural Sheaf Diffusion: A Symmetric Simplicial Set Framework for Higher-Order LearningComments: This manuscript has been submitted to IEEE Access for publicationSubjects: Machine Learning (cs.LG); Algebraic Topology (math.AT)
The absence of intrinsic adjacency relations and orientation systems in hypergraphs creates fundamental challenges for constructing sheaf Laplacians of arbitrary degrees. We resolve these limitations through symmetric simplicial sets derived directly from hypergraphs, which encode all possible oriented subrelations within each hyperedge as ordered tuples. This construction canonically defines adjacency via facet maps while inherently preserving hyperedge provenance. We establish that the normalized degree zero sheaf Laplacian on our induced symmetric simplicial set reduces exactly to the traditional graph normalized sheaf Laplacian when restricted to graphs, validating its mathematical consistency with prior graph-based sheaf theory. Furthermore, the induced structure preserves all structural information from the original hypergraph, ensuring that every multi-way relational detail is faithfully retained. Leveraging this framework, we introduce Hypergraph Neural Sheaf Diffusion (HNSD), the first principled extension of Neural Sheaf Diffusion (NSD) to hypergraphs. HNSD operates via normalized degree zero sheaf Laplacians over symmetric simplicial sets, resolving orientation ambiguity and adjacency sparsity inherent to hypergraph learning. Experimental evaluations demonstrate HNSD's competitive performance across established benchmarks.
- [555] arXiv:2505.07004 (replaced) [pdf, other]
-
Title: GuidedQuant: Large Language Model Quantization via Exploiting End Loss GuidanceJinuk Kim, Marwa El Halabi, Wonpyo Park, Clemens JS Schaefer, Deokjae Lee, Yeonhong Park, Jae W. Lee, Hyun Oh SongComments: ICML 2025Subjects: Machine Learning (cs.LG)
Post-training quantization is a key technique for reducing the memory and inference latency of large language models by quantizing weights and activations without requiring retraining. However, existing methods either (1) fail to account for the varying importance of hidden features to the end loss or, when incorporating end loss, (2) neglect the critical interactions between model weights. To address these limitations, we propose GuidedQuant, a novel quantization approach that integrates gradient information from the end loss into the quantization objective while preserving cross-weight dependencies within output channels. GuidedQuant consistently boosts the performance of state-of-the-art quantization methods across weight-only scalar, weight-only vector, and weight-and-activation quantization. Additionally, we introduce a novel non-uniform scalar quantization algorithm, which is guaranteed to monotonically decrease the quantization objective value, and outperforms existing methods in this category. We release the code at this https URL.
- [556] arXiv:2505.08740 (replaced) [pdf, html, other]
-
Title: Sensitivity-Constrained Fourier Neural Operators for Forward and Inverse Problems in Parametric Differential EquationsComments: Accepted and presented at ICLR 2025Subjects: Machine Learning (cs.LG); Computational Engineering, Finance, and Science (cs.CE)
Parametric differential equations of the form du/dt = f(u, x, t, p) are fundamental in science and engineering. While deep learning frameworks such as the Fourier Neural Operator (FNO) can efficiently approximate solutions, they struggle with inverse problems, sensitivity estimation (du/dp), and concept drift. We address these limitations by introducing a sensitivity-based regularization strategy, called Sensitivity-Constrained Fourier Neural Operators (SC-FNO). SC-FNO achieves high accuracy in predicting solution paths and consistently outperforms standard FNO and FNO with physics-informed regularization. It improves performance in parameter inversion tasks, scales to high-dimensional parameter spaces (tested with up to 82 parameters), and reduces both data and training requirements. These gains are achieved with a modest increase in training time (30% to 130% per epoch) and generalize across various types of differential equations and neural operators. Code and selected experiments are available at: this https URL
- [557] arXiv:2505.10930 (replaced) [pdf, html, other]
-
Title: Physics-informed Temporal Alignment for Auto-regressive PDE Foundation ModelsComments: Accepted as a conference paper in ICML2025Subjects: Machine Learning (cs.LG)
Auto-regressive partial differential equation (PDE) foundation models have shown great potential in handling time-dependent data. However, these models suffer from the shortcut problem deeply rooted in auto-regressive prediction, causing error accumulation. The challenge becomes particularly evident for out-of-distribution data, as the pretraining performance may approach random model initialization for downstream tasks with long-term dynamics. To deal with this problem, we propose physics-informed temporal alignment (PITA), a self-supervised learning framework inspired by inverse problem solving. Specifically, PITA aligns the physical dynamics discovered at different time steps on each given PDE trajectory by integrating physics-informed constraints into the self-supervision signal. The alignment is derived from observation data without relying on known physics priors, indicating strong generalization ability to the out-of-distribution data. Extensive experiments show that PITA significantly enhances the accuracy and robustness of existing foundation models on diverse time-dependent PDE data. The code is available at this https URL.
- [558] arXiv:2505.11197 (replaced) [pdf, html, other]
-
Title: Modeling Cell Dynamics and Interactions with Unbalanced Mean Field Schrödinger BridgeSubjects: Machine Learning (cs.LG); Optimization and Control (math.OC); Quantitative Methods (q-bio.QM)
Modeling the dynamics from sparsely time-resolved snapshot data is crucial for understanding complex cellular processes and behavior. Existing methods leverage optimal transport, Schrödinger bridge theory, or their variants to simultaneously infer stochastic, unbalanced dynamics from snapshot data. However, these approaches remain limited in their ability to account for cell-cell interactions. This integration is essential in real-world scenarios since intercellular communications are fundamental life processes and can influence cell state-transition dynamics. To address this challenge, we formulate the Unbalanced Mean-Field Schrödinger Bridge (UMFSB) framework to model unbalanced stochastic interaction dynamics from snapshot data. Inspired by this framework, we further propose CytoBridge, a deep learning algorithm designed to approximate the UMFSB problem. By explicitly modeling cellular transitions, proliferation, and interactions through neural networks, CytoBridge offers the flexibility to learn these processes directly from data. The effectiveness of our method has been extensively validated using both synthetic gene regulatory data and real scRNA-seq datasets. Compared to existing methods, CytoBridge identifies growth, transition, and interaction patterns, eliminates false transitions, and reconstructs the developmental landscape with greater accuracy.
- [559] arXiv:2505.11396 (replaced) [pdf, html, other]
-
Title: Finding Counterfactual Evidences for Node ClassificationComments: Accepted by KDD 2025Subjects: Machine Learning (cs.LG); Databases (cs.DB)
Counterfactual learning is emerging as an important paradigm, rooted in causality, which promises to alleviate common issues of graph neural networks (GNNs), such as fairness and interpretability. However, as in many real-world application domains where conducting randomized controlled trials is impractical, one has to rely on available observational (factual) data to detect counterfactuals. In this paper, we introduce and tackle the problem of searching for counterfactual evidences for the GNN-based node classification task. A counterfactual evidence is a pair of nodes such that, regardless they exhibit great similarity both in the features and in their neighborhood subgraph structures, they are classified differently by the GNN. We develop effective and efficient search algorithms and a novel indexing solution that leverages both node features and structural information to identify counterfactual evidences, and generalizes beyond any specific GNN. Through various downstream applications, we demonstrate the potential of counterfactual evidences to enhance fairness and accuracy of GNNs.
- [560] arXiv:2505.16401 (replaced) [pdf, other]
-
Title: Divide-Fuse-Conquer: Eliciting "Aha Moments" in Multi-Scenario GamesComments: 25 pages, 13 figures, and 8 tablesSubjects: Machine Learning (cs.LG)
Large language models (LLMs) have been observed to suddenly exhibit advanced reasoning abilities during reinforcement learning (RL), resembling an ``aha moment'' triggered by simple outcome-based rewards. While RL has proven effective in eliciting such breakthroughs in tasks involving mathematics, coding, and vision, it faces significant challenges in multi-scenario games. The diversity of game rules, interaction modes, and environmental complexities often leads to policies that perform well in one scenario but fail to generalize to others. Simply combining multiple scenarios during training introduces additional challenges, such as training instability and poor performance. To overcome these challenges, we propose Divide-Fuse-Conquer, a framework designed to enhance generalization in multi-scenario RL. This approach starts by heuristically grouping games based on characteristics such as rules and difficulties. Specialized models are then trained for each group to excel at games in the group is what we refer to as the divide step. Next, we fuse model parameters from different groups as a new model, and continue training it for multiple groups, until the scenarios in all groups are conquered. Experiments across 18 TextArena games show that Qwen2.5-32B-Align trained with the Divide-Fuse-Conquer strategy reaches a performance level comparable to Claude3.5, achieving 7 wins and 4 draws. We hope our approach can inspire future research on using reinforcement learning to improve the generalization of LLMs.
- [561] arXiv:2505.17155 (replaced) [pdf, html, other]
-
Title: TrimR: Verifier-based Training-Free Thinking Compression for Efficient Test-Time ScalingWeizhe Lin, Xing Li, Zhiyuan Yang, Xiaojin Fu, Hui-Ling Zhen, Yaoyuan Wang, Xianzhi Yu, Wulong Liu, Xiaosong Li, Mingxuan YuanSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Large Reasoning Models (LRMs) demonstrate exceptional capability in tackling complex mathematical, logical, and coding tasks by leveraging extended Chain-of-Thought (CoT) reasoning. Test-time scaling methods, such as prolonging CoT with explicit token-level exploration, can push LRMs' accuracy boundaries, but they incur significant decoding overhead. A key inefficiency source is LRMs often generate redundant thinking CoTs, which demonstrate clear structured overthinking and underthinking patterns. Inspired by human cognitive reasoning processes and numerical optimization theories, we propose TrimR, a verifier-based, training-free, efficient framework for dynamic CoT compression to trim reasoning and enhance test-time scaling, explicitly tailored for production-level deployment. Our method employs a lightweight, pretrained, instruction-tuned verifier to detect and truncate redundant intermediate thoughts of LRMs without any LRM or verifier fine-tuning. We present both the core algorithm and asynchronous online system engineered for high-throughput industrial applications. Empirical evaluations on Ascend NPUs and vLLM show that our framework delivers substantial gains in inference efficiency under large-batch workloads. In particular, on the four MATH500, AIME24, AIME25, and GPQA benchmarks, the reasoning runtime of Pangu Pro MoE, Pangu-R-38B, QwQ-32B, and DeepSeek-R1-Distill-Qwen-32B is improved by up to 70% with negligible impact on accuracy.
- [562] arXiv:2505.17370 (replaced) [pdf, html, other]
-
Title: FRIREN: Beyond Trajectories -- A Spectral Lens on TimeComments: 37 pages, 4 figures. Submitted to NeurIPS 2025. Public code at this https URLSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Long-term time-series forecasting (LTSF) models are often presented as general-purpose solutions that can be applied across domains, implicitly assuming that all data is pointwise predictable. Using chaotic systems such as Lorenz-63 as a case study, we argue that geometric structure - not pointwise prediction - is the right abstraction for a dynamic-agnostic foundational model. Minimizing the Wasserstein-2 distance (W2), which captures geometric changes, and providing a spectral view of dynamics are essential for long-horizon forecasting. Our model, FRIREN (Flow-inspired Representations via Interpretable Eigen-networks), implements an augmented normalizing-flow block that embeds data into a normally distributed latent representation. It then generates a W2-efficient optimal path that can be decomposed into rotation, scaling, inverse rotation, and translation. This architecture yields locally generated, geometry-preserving predictions that are independent of the underlying dynamics, and a global spectral representation that functions as a finite Koopman operator with a small modification. This enables practitioners to identify which modes grow, decay, or oscillate, both locally and system-wide. FRIREN achieves an MSE of 11.4, MAE of 1.6, and SWD of 0.96 on Lorenz-63 in a 336-in, 336-out, dt=0.01 setting, surpassing TimeMixer (MSE 27.3, MAE 2.8, SWD 2.1). The model maintains effective prediction for 274 out of 336 steps, approximately 2.5 Lyapunov times. On Rossler (96-in, 336-out), FRIREN achieves an MSE of 0.0349, MAE of 0.0953, and SWD of 0.0170, outperforming TimeMixer's MSE of 4.3988, MAE of 0.886, and SWD of 3.2065. FRIREN is also competitive on standard LTSF datasets such as ETT and Weather. By connecting modern generative flows with classical spectral analysis, FRIREN makes long-term forecasting both accurate and interpretable, setting a new benchmark for LTSF model design.
- [563] arXiv:2505.17371 (replaced) [pdf, html, other]
-
Title: An End-to-End Approach for Child Reading Assessment in the Xhosa LanguageSergio Chevtchenko, Nikhil Navas, Rafaella Vale, Franco Ubaudi, Sipumelele Lucwaba, Cally Ardington, Soheil Afshar, Mark Antoniou, Saeed AfsharComments: Paper accepted on AIED 2025 containing 14 pages, 6 figures and 4 tablesSubjects: Machine Learning (cs.LG); Computation and Language (cs.CL)
Child literacy is a strong predictor of life outcomes at the subsequent stages of an individual's life. This points to a need for targeted interventions in vulnerable low and middle income populations to help bridge the gap between literacy levels in these regions and high income ones. In this effort, reading assessments provide an important tool to measure the effectiveness of these programs and AI can be a reliable and economical tool to support educators with this task. Developing accurate automatic reading assessment systems for child speech in low-resource languages poses significant challenges due to limited data and the unique acoustic properties of children's voices. This study focuses on Xhosa, a language spoken in South Africa, to advance child speech recognition capabilities. We present a novel dataset composed of child speech samples in Xhosa. The dataset is available upon request and contains ten words and letters, which are part of the Early Grade Reading Assessment (EGRA) system. Each recording is labeled with an online and cost-effective approach by multiple markers and a subsample is validated by an independent EGRA reviewer. This dataset is evaluated with three fine-tuned state-of-the-art end-to-end models: wav2vec 2.0, HuBERT, and Whisper. The results indicate that the performance of these models can be significantly influenced by the amount and balancing of the available training data, which is fundamental for cost-effective large dataset collection. Furthermore, our experiments indicate that the wav2vec 2.0 performance is improved by training on multiple classes at a time, even when the number of available samples is constrained.
- [564] arXiv:2505.17662 (replaced) [pdf, html, other]
-
Title: Automating Versatile Time-Series Analysis with Tiny Transformers on Embedded FPGAsComments: 6 pages, 5 figures, 1 table, accepted by IEEE Computer Society Annual Symposium on VLSI (ISVLSI 2025)Subjects: Machine Learning (cs.LG)
Transformer-based models have shown strong performance across diverse time-series tasks, but their deployment on resource-constrained devices remains challenging due to high memory and computational demand. While prior work targeting Microcontroller Units (MCUs) has explored hardware-specific optimizations, such approaches are often task-specific and limited to 8-bit fixed-point precision. Field-Programmable Gate Arrays (FPGAs) offer greater flexibility, enabling fine-grained control over data precision and architecture. However, existing FPGA-based deployments of Transformers for time-series analysis typically focus on high-density platforms with manual configuration. This paper presents a unified and fully automated deployment framework for Tiny Transformers on embedded FPGAs. Our framework supports a compact encoder-only Transformer architecture across three representative time-series tasks (forecasting, classification, and anomaly detection). It combines quantization-aware training (down to 4 bits), hardware-aware hyperparameter search using Optuna, and automatic VHDL generation for seamless deployment. We evaluate our framework on six public datasets across two embedded FPGA platforms. Results show that our framework produces integer-only, task-specific Transformer accelerators achieving as low as 0.033 mJ per inference with millisecond latency on AMD Spartan-7, while also providing insights into deployment feasibility on Lattice iCE40. All source code will be released in the GitHub repository (this https URL).
- [565] arXiv:2505.19014 (replaced) [pdf, html, other]
-
Title: Tokenizing Electron Cloud in Protein-Ligand Interaction LearningHaitao Lin, Odin Zhang, Jia Xu, Yunfan Liu, Zheng Cheng, Lirong Wu, Yufei Huang, Zhifeng Gao, Stan Z. LiComments: conference paperSubjects: Machine Learning (cs.LG); Chemical Physics (physics.chem-ph); Quantitative Methods (q-bio.QM)
The affinity and specificity of protein-molecule binding directly impact functional outcomes, uncovering the mechanisms underlying biological regulation and signal transduction. Most deep-learning-based prediction approaches focus on structures of atoms or fragments. However, quantum chemical properties, such as electronic structures, are the key to unveiling interaction patterns but remain largely underexplored. To bridge this gap, we propose ECBind, a method for tokenizing electron cloud signals into quantized embeddings, enabling their integration into downstream tasks such as binding affinity prediction. By incorporating electron densities, ECBind helps uncover binding modes that cannot be fully represented by atom-level models. Specifically, to remove the redundancy inherent in electron cloud signals, a structure-aware transformer and hierarchical codebooks encode 3D binding sites enriched with electron structures into tokens. These tokenized codes are then used for specific tasks with labels. To extend its applicability to a wider range of scenarios, we utilize knowledge distillation to develop an electron-cloud-agnostic prediction model. Experimentally, ECBind demonstrates state-of-the-art performance across multiple tasks, achieving improvements of 6.42\% and 15.58\% in per-structure Pearson and Spearman correlation coefficients, respectively.
- [566] arXiv:2505.19193 (replaced) [pdf, html, other]
-
Title: Interpretable Graph Learning Over Sets of Temporally-Sparse DataAndrea Zerio, Maya Bechler-Speicher, Maor Huri, Marie Vibeke Vestergaard, Ran Gilad-Bachrach, Tine Jess, Samir Bhatt, Aleksejs SazonovsSubjects: Machine Learning (cs.LG)
Real-world medical data often includes measurements from multiple signals that are collected at irregular and asynchronous time intervals. For example, different types of blood tests can be measured at different times and frequencies, resulting in fragmented and unevenly scattered temporal data. Similar issues of irregular sampling of different attributes occur in other domains, such as monitoring of large systems using event log files or the spread of fake news on social networks. Effectively learning from such data requires models that can handle sets of temporally sparse and heterogeneous signals. In this paper, we propose Graph Mixing Additive Networks (GMAN), a novel and interpretable-by-design model for learning over irregular sets of temporal signals. Our method achieves state-of-the-art performance in real-world medical tasks, including a 4-point increase in the AUROC score of in-hospital mortality prediction, compared to existing methods. We further showcase GMAN's flexibility by applying it to a fake news detection task. We demonstrate how its interpretability capabilities, including node-level, graph-level, and subset-level importance, allow for transition phases detection and gaining medical insights with real-world high-stakes implications. Finally, we provide theoretical insights on GMAN expressive power.
- [567] arXiv:2505.19433 (replaced) [pdf, html, other]
-
Title: Can Compressed LLMs Truly Act? An Empirical Evaluation of Agentic Capabilities in LLM CompressionComments: Accepted by ICML2025 as PosterSubjects: Machine Learning (cs.LG)
Post-training compression reduces the computational and memory costs of large language models (LLMs), enabling resource-efficient deployment. However, existing compression benchmarks only focus on language modeling (e.g., perplexity) and natural language understanding tasks (e.g., GLUE accuracy), ignoring the agentic capabilities - workflow, tool use/function call, long-context understanding and real-world application. We introduce the Agent Compression Benchmark (ACBench), the first comprehensive benchmark for evaluating how compression impacts LLMs' agentic abilities. ACBench spans (1) 12 tasks across 4 capabilities (e.g., WorfBench for workflow generation, Needle-in-Haystack for long-context retrieval), (2) quantization (GPTQ, AWQ) and pruning (Wanda, SparseGPT), and (3) 15 models, including small (Gemma-2B), standard (Qwen2.5 7B-32B), and distilled reasoning LLMs (DeepSeek-R1-Distill). Our experiments reveal compression tradeoffs: 4-bit quantization preserves workflow generation and tool use (1%-3% drop) but degrades real-world application accuracy by 10%-15%. We introduce ERank, Top-k Ranking Correlation and Energy to systematize analysis. ACBench provides actionable insights for optimizing LLM compression in agentic scenarios. The code can be found in this https URL.
- [568] arXiv:2505.19552 (replaced) [pdf, html, other]
-
Title: On scalable and efficient training of diffusion samplersSubjects: Machine Learning (cs.LG)
We address the challenge of training diffusion models to sample from unnormalized energy distributions in the absence of data, the so-called diffusion samplers. Although these approaches have shown promise, they struggle to scale in more demanding scenarios where energy evaluations are expensive and the sampling space is high-dimensional. To address this limitation, we propose a scalable and sample-efficient framework that properly harmonizes the powerful classical sampling method and the diffusion sampler. Specifically, we utilize Monte Carlo Markov chain (MCMC) samplers with a novelty-based auxiliary energy as a Searcher to collect off-policy samples, using an auxiliary energy function to compensate for exploring modes the diffusion sampler rarely visits. These off-policy samples are then combined with on-policy data to train the diffusion sampler, thereby expanding its coverage of the energy landscape. Furthermore, we identify primacy bias, i.e., the preference of samplers for early experience during training, as the main cause of mode collapse during training, and introduce a periodic re-initialization trick to resolve this issue. Our method significantly improves sample efficiency on standard benchmarks for diffusion samplers and also excels at higher-dimensional problems and real-world molecular conformer generation.
- [569] arXiv:2505.19669 (replaced) [pdf, html, other]
-
Title: Zero-Shot Streaming Text to Speech Synthesis with Transducer and Auto-Regressive ModelingHaiyang Sun, Shujie Hu, Shujie Liu, Lingwei Meng, Hui Wang, Bing Han, Yifan Yang, Yanqing Liu, Sheng Zhao, Yan Lu, Yanmin QianSubjects: Machine Learning (cs.LG)
Zero-shot streaming text-to-speech is an important research topic in human-computer interaction. Existing methods primarily use a lookahead mechanism, relying on future text to achieve natural streaming speech synthesis, which introduces high processing latency. To address this issue, we propose SMLLE, a streaming framework for generating high-quality speech frame-by-frame. SMLLE employs a Transducer to convert text into semantic tokens in real time while simultaneously obtaining duration alignment information. The combined outputs are then fed into a fully autoregressive (AR) streaming model to reconstruct mel-spectrograms. To further stabilize the generation process, we design a Delete < Bos > Mechanism that allows the AR model to access future text introducing as minimal delay as possible. Experimental results suggest that the SMLLE outperforms current streaming TTS methods and achieves comparable performance over sentence-level TTS systems. Samples are available on this http URL.
- [570] arXiv:2505.19946 (replaced) [pdf, html, other]
-
Title: Inverse Q-Learning Done Right: Offline Imitation Learning in $Q^π$-Realizable MDPsSubjects: Machine Learning (cs.LG)
We study the problem of offline imitation learning in Markov decision processes (MDPs), where the goal is to learn a well-performing policy given a dataset of state-action pairs generated by an expert policy. Complementing a recent line of work on this topic that assumes the expert belongs to a tractable class of known policies, we approach this problem from a new angle and leverage a different type of structural assumption about the environment. Specifically, for the class of linear $Q^\pi$-realizable MDPs, we introduce a new algorithm called saddle-point offline imitation learning (\SPOIL), which is guaranteed to match the performance of any expert up to an additive error $\varepsilon$ with access to $\mathcal{O}(\varepsilon^{-2})$ samples. Moreover, we extend this result to possibly non-linear $Q^\pi$-realizable MDPs at the cost of a worse sample complexity of order $\mathcal{O}(\varepsilon^{-4})$. Finally, our analysis suggests a new loss function for training critic networks from expert data in deep imitation learning. Empirical evaluations on standard benchmarks demonstrate that the neural net implementation of \SPOIL is superior to behavior cloning and competitive with state-of-the-art algorithms.
- [571] arXiv:2505.20734 (replaced) [pdf, html, other]
-
Title: Adversarial bandit optimization for approximately linear functionsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
We consider a bandit optimization problem for nonconvex and non-smooth functions, where in each trial the loss function is the sum of a linear function and a small but arbitrary perturbation chosen after observing the player's choice. We give both expected and high probability regret bounds for the problem. Our result also implies an improved high-probability regret bound for the bandit linear optimization, a special case with no perturbation. We also give a lower bound on the expected regret.
- [572] arXiv:2505.20896 (replaced) [pdf, html, other]
-
Title: How Do Transformers Learn Variable Binding in Symbolic Programs?Comments: 16 pages, 10 figures, 1 table. To appear in the Proceedings of the 42nd International Conference on Machine Learning (ICML 2025). v2: Added link to Variable Scope in abstractSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
Variable binding -- the ability to associate variables with values -- is fundamental to symbolic computation and cognition. Although classical architectures typically implement variable binding via addressable memory, it is not well understood how modern neural networks lacking built-in binding operations may acquire this capacity. We investigate this by training a Transformer to dereference queried variables in symbolic programs where variables are assigned either numerical constants or other variables. Each program requires following chains of variable assignments up to four steps deep to find the queried value, and also contains irrelevant chains of assignments acting as distractors. Our analysis reveals a developmental trajectory with three distinct phases during training: (1) random prediction of numerical constants, (2) a shallow heuristic prioritizing early variable assignments, and (3) the emergence of a systematic mechanism for dereferencing assignment chains. Using causal interventions, we find that the model learns to exploit the residual stream as an addressable memory space, with specialized attention heads routing information across token positions. This mechanism allows the model to dynamically track variable bindings across layers, resulting in accurate dereferencing. Our results show how Transformer models can learn to implement systematic variable binding without explicit architectural support, bridging connectionist and symbolic approaches. To facilitate reproducible research, we developed Variable Scope, an interactive web platform for exploring our findings at this https URL
- [573] arXiv:2505.21119 (replaced) [pdf, html, other]
-
Title: Universal Value-Function UncertaintiesMoritz A. Zanger, Max Weltevrede, Yaniv Oren, Pascal R. Van der Vaart, Caroline Horsch, Wendelin Böhmer, Matthijs T. J. SpaanSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
Estimating epistemic uncertainty in value functions is a crucial challenge for many aspects of reinforcement learning (RL), including efficient exploration, safe decision-making, and offline RL. While deep ensembles provide a robust method for quantifying value uncertainty, they come with significant computational overhead. Single-model methods, while computationally favorable, often rely on heuristics and typically require additional propagation mechanisms for myopic uncertainty estimates. In this work we introduce universal value-function uncertainties (UVU), which, similar in spirit to random network distillation (RND), quantify uncertainty as squared prediction errors between an online learner and a fixed, randomly initialized target network. Unlike RND, UVU errors reflect policy-conditional value uncertainty, incorporating the future uncertainties any given policy may encounter. This is due to the training procedure employed in UVU: the online network is trained using temporal difference learning with a synthetic reward derived from the fixed, randomly initialized target network. We provide an extensive theoretical analysis of our approach using neural tangent kernel (NTK) theory and show that in the limit of infinite network width, UVU errors are exactly equivalent to the variance of an ensemble of independent universal value functions. Empirically, we show that UVU achieves equal performance to large ensembles on challenging multi-task offline RL settings, while offering simplicity and substantial computational savings.
- [574] arXiv:2505.21938 (replaced) [pdf, html, other]
-
Title: Practical Adversarial Attacks on Stochastic Bandits via Fake Data InjectionSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
Adversarial attacks on stochastic bandits have traditionally relied on some unrealistic assumptions, such as per-round reward manipulation and unbounded perturbations, limiting their relevance to real-world systems. We propose a more practical threat model, Fake Data Injection, which reflects realistic adversarial constraints: the attacker can inject only a limited number of bounded fake feedback samples into the learner's history, simulating legitimate interactions. We design efficient attack strategies under this model, explicitly addressing both magnitude constraints (on reward values) and temporal constraints (on when and how often data can be injected). Our theoretical analysis shows that these attacks can mislead both Upper Confidence Bound (UCB) and Thompson Sampling algorithms into selecting a target arm in nearly all rounds while incurring only sublinear attack cost. Experiments on synthetic and real-world datasets validate the effectiveness of our strategies, revealing significant vulnerabilities in widely used stochastic bandit algorithms under practical adversarial scenarios.
- [575] arXiv:2505.22984 (replaced) [pdf, html, other]
-
Title: A Computational Approach to Improving Fairness in K-means ClusteringComments: 14 pages, 5 figuresSubjects: Machine Learning (cs.LG); Computers and Society (cs.CY)
The popular K-means clustering algorithm potentially suffers from a major weakness for further analysis or interpretation. Some cluster may have disproportionately more (or fewer) points from one of the subpopulations in terms of some sensitive variable, e.g., gender or race. Such a fairness issue may cause bias and unexpected social consequences. This work attempts to improve the fairness of K-means clustering with a two-stage optimization formulation--clustering first and then adjust cluster membership of a small subset of selected data points. Two computationally efficient algorithms are proposed in identifying those data points that are expensive for fairness, with one focusing on nearest data points outside of a cluster and the other on highly 'mixed' data points. Experiments on benchmark datasets show substantial improvement on fairness with a minimal impact to clustering quality. The proposed algorithms can be easily extended to a broad class of clustering algorithms or fairness metrics.
- [576] arXiv:2505.23337 (replaced) [pdf, html, other]
-
Title: Matryoshka Model Learning for Improved Elastic Student ModelsChetan Verma, Aditya Srinivas Timmaraju, Cho-Jui Hsieh, Suyash Damle, Ngot Bui, Yang Zhang, Wen Chen, Xin Liu, Prateek Jain, Inderjit S DhillonComments: 10 pages, 5 figures, Accepted at KDD 2025Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Industry-grade ML models are carefully designed to meet rapidly evolving serving constraints, which requires significant resources for model development. In this paper, we propose MatTA, a framework for training multiple accurate Student models using a novel Teacher-TA-Student recipe. TA models are larger versions of the Student models with higher capacity, and thus allow Student models to better relate to the Teacher model and also bring in more domain-specific expertise. Furthermore, multiple accurate Student models can be extracted from the TA model. Therefore, despite only one training run, our methodology provides multiple servable options to trade off accuracy for lower serving cost. We demonstrate the proposed method, MatTA, on proprietary datasets and models. Its practical efficacy is underscored by live A/B tests within a production ML system, demonstrating 20% improvement on a key metric. We also demonstrate our method on GPT-2 Medium, a public model, and achieve relative improvements of over 24% on SAT Math and over 10% on the LAMBADA benchmark.
- [577] arXiv:2505.23378 (replaced) [pdf, html, other]
-
Title: Meta-Learning Approaches for Speaker-Dependent Voice Fatigue ModelsComments: 5 pages, 3 figures. To appear at Interspeech 2025Subjects: Machine Learning (cs.LG)
Speaker-dependent modelling can substantially improve performance in speech-based health monitoring applications. While mixed-effect models are commonly used for such speaker adaptation, they require computationally expensive retraining for each new observation, making them impractical in a production environment. We reformulate this task as a meta-learning problem and explore three approaches of increasing complexity: ensemble-based distance models, prototypical networks, and transformer-based sequence models. Using pre-trained speech embeddings, we evaluate these methods on a large longitudinal dataset of shift workers (N=1,185, 10,286 recordings), predicting time since sleep from speech as a function of fatigue, a symptom commonly associated with ill-health. Our results demonstrate that all meta-learning approaches tested outperformed both cross-sectional and conventional mixed-effects models, with a transformer-based method achieving the strongest performance.
- [578] arXiv:2505.23527 (replaced) [pdf, html, other]
-
Title: Normalizing Flows are Capable Models for RLComments: Project page with code - this https URLSubjects: Machine Learning (cs.LG)
Modern reinforcement learning (RL) algorithms have found success by using powerful probabilistic models, such as transformers, energy-based models, and diffusion/flow-based models. To this end, RL researchers often choose to pay the price of accommodating these models into their algorithms -- diffusion models are expressive, but are computationally intensive due to their reliance on solving differential equations, while autoregressive transformer models are scalable but typically require learning discrete representations. Normalizing flows (NFs), by contrast, seem to provide an appealing alternative, as they enable likelihoods and sampling without solving differential equations or autoregressive architectures. However, their potential in RL has received limited attention, partly due to the prevailing belief that normalizing flows lack sufficient expressivity. We show that this is not the case. Building on recent work in NFs, we propose a single NF architecture which integrates seamlessly into RL algorithms, serving as a policy, Q-function, and occupancy measure. Our approach leads to much simpler algorithms, and achieves higher performance in imitation learning, offline, goal conditioned RL and unsupervised RL.
- [579] arXiv:2505.24034 (replaced) [pdf, html, other]
-
Title: LlamaRL: A Distributed Asynchronous Reinforcement Learning Framework for Efficient Large-scale LLM TrainingBo Wu, Sid Wang, Yunhao Tang, Jia Ding, Eryk Helenowski, Liang Tan, Tengyu Xu, Tushar Gowda, Zhengxing Chen, Chen Zhu, Xiaocheng Tang, Yundi Qian, Beibei Zhu, Rui HouSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
Reinforcement Learning (RL) has become the most effective post-training approach for improving the capabilities of Large Language Models (LLMs). In practice, because of the high demands on latency and memory, it is particularly challenging to develop an efficient RL framework that reliably manages policy models with hundreds to thousands of billions of parameters.
In this paper, we present LlamaRL, a fully distributed, asynchronous RL framework optimized for efficient training of large-scale LLMs with various model sizes (8B, 70B, and 405B parameters) on GPU clusters ranging from a handful to thousands of devices. LlamaRL introduces a streamlined, single-controller architecture built entirely on native PyTorch, enabling modularity, ease of use, and seamless scalability to thousands of GPUs. We also provide a theoretical analysis of LlamaRL's efficiency, including a formal proof that its asynchronous design leads to strict RL speed-up. Empirically during the Llama 3 post-training, by leveraging best practices such as colocated model offloading, asynchronous off-policy training, and distributed direct memory access for weight synchronization, LlamaRL achieves significant efficiency gains -- up to 10.7x speed-up compared to DeepSpeed-Chat-like systems on a 405B-parameter policy model. Furthermore, the efficiency advantage continues to grow with increasing model scale, demonstrating the framework's suitability for future large-scale RL training. - [580] arXiv:2505.24584 (replaced) [pdf, html, other]
-
Title: AutoChemSchematic AI: A Closed-Loop, Physics-Aware Agentic Framework for Auto-Generating Chemical Process and Instrumentation DiagramsSubjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
Recent advancements in generative AI have accelerated the discovery of novel chemicals and materials; however, transitioning these discoveries to industrial-scale production remains a critical bottleneck, as it requires the development of entirely new chemical manufacturing processes. Current AI methods cannot auto-generate PFDs or PIDs, despite their critical role in scaling chemical processes, while adhering to engineering constraints. We present a closed loop, physics aware framework for the automated generation of industrially viable PFDs and PIDs. The framework integrates domain specialized small scale language models (SLMs) (trained for chemical process QA tasks) with first principles simulation, leveraging three key components: (1) a hierarchical knowledge graph of process flow and instrumentation descriptions for 1,020+ chemicals, (2) a multi-stage training pipeline that fine tunes domain specialized SLMs on synthetic datasets via Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Retrieval-Augmented Instruction Tuning (RAIT), and (3) DWSIM based simulator in the loop validation to ensure feasibility. To improve both runtime efficiency and model compactness, the framework incorporates advanced inference time optimizations including FlashAttention, Lookahead Decoding, PagedAttention with KV-cache quantization, and Test Time Inference Scaling and independently applies structural pruning techniques (width and depth) guided by importance heuristics to reduce model size with minimal accuracy loss. Experiments demonstrate that the framework generates simulator-validated process descriptions with high fidelity, outperforms baseline methods in correctness, and generalizes to unseen chemicals. By bridging AI-driven design with industrial-scale feasibility, this work significantly reduces R&D timelines from lab discovery to plant deployment.
- [581] arXiv:2207.06418 (replaced) [pdf, html, other]
-
Title: Open High-Resolution Satellite Imagery: The WorldStrat Dataset -- With Application to Super-ResolutionComments: Published in 36th Conference on Neural Information Processing Systems (NeurIPS 2022) Track on Datasets and BenchmarksSubjects: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Applications (stat.AP)
Analyzing the planet at scale with satellite imagery and machine learning is a dream that has been constantly hindered by the cost of difficult-to-access highly-representative high-resolution imagery. To remediate this, we introduce here the WorldStrat dataset. The largest and most varied such publicly available dataset, at Airbus SPOT 6/7 satellites' high resolution of up to 1.5 m/pixel, empowered by European Space Agency's Phi-Lab as part of the ESA-funded QueryPlanet project, we curate nearly 10,000 sqkm of unique locations to ensure stratified representation of all types of land-use across the world: from agriculture to ice caps, from forests to multiple urbanization densities. We also enrich those with locations typically under-represented in ML datasets: sites of humanitarian interest, illegal mining sites, and settlements of persons at risk. We temporally-match each high-resolution image with multiple low-resolution images from the freely accessible lower-resolution Sentinel-2 satellites at 10 m/pixel. We accompany this dataset with an open-source Python package to: rebuild or extend the WorldStrat dataset, train and infer baseline algorithms, and learn with abundant tutorials, all compatible with the popular EO-learn toolbox. We hereby hope to foster broad-spectrum applications of ML to satellite imagery, and possibly develop from free public low-resolution Sentinel2 imagery the same power of analysis allowed by costly private high-resolution imagery. We illustrate this specific point by training and releasing several highly compute-efficient baselines on the task of Multi-Frame Super-Resolution. High-resolution Airbus imagery is CC BY-NC, while the labels and Sentinel2 imagery are CC BY, and the source code and pre-trained models under BSD. The dataset is available at this https URL and the software package at this https URL .
- [582] arXiv:2212.05260 (replaced) [pdf, html, other]
-
Title: Examining marginal properness in the external validation of survival models with squared and logarithmic lossesSubjects: Statistics Theory (math.ST); Machine Learning (cs.LG); Applications (stat.AP)
Scoring rules promote rational and honest decision-making, which is important for model evaluation and becoming increasingly important for automated procedures such as `AutoML'. In this paper we survey common squared and logarithmic scoring rules for survival analysis, with a focus on their theoretical and empirical properness. We introduce a marginal definition of properness and show that both the Integrated Survival Brier Score (ISBS) and the Right-Censored Log-Likelihood (RCLL) are theoretically improper under this definition. We also investigate a new class of losses that may inform future survival scoring rules. Simulation experiments reveal that both the ISBS and RCLL behave as proper scoring rules in practice. The RCLL showed no violations across all settings, while ISBS exhibited only minor, negligible violations at extremely small sample sizes, suggesting one can trust results from historical experiments. As such we advocate for both the RCLL and ISBS in external validation of models, including in automated procedures. However, we note practical challenges in estimating these losses including estimation of censoring distributions and densities; as such further research is required to advance development of robust and honest evaluation in survival analysis.
- [583] arXiv:2306.17329 (replaced) [pdf, html, other]
-
Title: Kernel $ε$-Greedy for Multi-Armed Bandits with CovariatesSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST)
We consider the $\epsilon$-greedy strategy for the multi-arm bandit with covariates (MABC) problem, where the mean reward functions are assumed to lie in a reproducing kernel Hilbert space (RKHS). We propose to estimate the unknown mean reward functions using an online weighted kernel ridge regression estimator, and show the resultant estimator to be consistent under appropriate decay rates of the exploration probability sequence, $\{\epsilon_t\}_t$, and regularization parameter, $\{\lambda_t\}_t$. Moreover, we show that for any choice of kernel and the corresponding RKHS, we achieve a sub-linear regret rate depending on the intrinsic dimensionality of the RKHS. Furthermore, we achieve the optimal regret rate of $\sqrt{T}$ under a margin condition for finite-dimensional RKHS.
- [584] arXiv:2307.00438 (replaced) [pdf, html, other]
-
Title: Towards Resource-Efficient Streaming of Large-Scale Medical Image Datasets for Deep LearningComments: 17 pages, 4 figures, 10 tables, accepted to MIDL'25Subjects: Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR); Machine Learning (cs.LG)
Large-scale medical imaging datasets have accelerated deep learning (DL) for medical image analysis. However, the large scale of these datasets poses a challenge for researchers, resulting in increased storage and bandwidth requirements for hosting and accessing them. Since different researchers have different use cases and require different resolutions or formats for DL, it is neither feasible to anticipate every researcher's needs nor practical to store data in multiple resolutions and formats. To that end, we propose the Medical Image Streaming Toolkit (MIST), a format-agnostic database that enables streaming of medical images at different resolutions and formats from a single high-resolution copy. We evaluated MIST across eight popular, large-scale medical imaging datasets spanning different body parts, modalities, and formats. Our results showed that our framework reduced the storage and bandwidth requirements for hosting and downloading datasets without impacting image quality. We demonstrate that MIST addresses the challenges posed by large-scale medical imaging datasets by building a data-efficient and format-agnostic database to meet the diverse needs of researchers and reduce barriers to DL research in medical imaging.
- [585] arXiv:2307.14634 (replaced) [pdf, html, other]
-
Title: Fact-Checking of AI-Generated ReportsComments: 10 pages, 3 figures, 3 tablesSubjects: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
With advances in generative artificial intelligence (AI), it is now possible to produce realistic-looking automated reports for preliminary reads of radiology images. This can expedite clinical workflows, improve accuracy and reduce overall costs. However, it is also well-known that such models often hallucinate, leading to false findings in the generated reports. In this paper, we propose a new method of fact-checking of AI-generated reports using their associated images. Specifically, the developed examiner differentiates real and fake sentences in reports by learning the association between an image and sentences describing real or potentially fake findings. To train such an examiner, we first created a new dataset of fake reports by perturbing the findings in the original ground truth radiology reports associated with images. Text encodings of real and fake sentences drawn from these reports are then paired with image encodings to learn the mapping to real/fake labels. The utility of such an examiner is demonstrated for verifying automatically generated reports by detecting and removing fake sentences. Future generative AI approaches can use the resulting tool to validate their reports leading to a more responsible use of AI in expediting clinical workflows.
- [586] arXiv:2308.01835 (replaced) [pdf, html, other]
-
Title: Resampled Confidence Regions with Exponential Shrinkage for the Regression Function of Binary ClassificationSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
The regression function is one of the key objects of binary classification, since it not only determines a Bayes optimal classifier, hence, defines an optimal decision boundary, but also encodes the conditional distribution of the output given the input. In this paper we build distribution-free confidence regions for the regression function for any user-chosen confidence level and any finite sample size based on a resampling test. These regions are abstract, as the model class can be almost arbitrary, e.g., it does not have to be finitely parameterized. We prove the strong uniform consistency of a new empirical risk minimization based approach for model classes with finite pseudo-dimensions and inverse Lipschitz parameterizations. We provide exponential probably approximately correct bounds on the $L_2$ sizes of these regions, and demonstrate the ideas on specific models. Additionally, we also consider a k-nearest neighbors based method, for which we prove strong pointwise bounds on the probability of exclusion. Finally, the constructions are illustrated on a logistic model class and compared to the asymptotic ellipsoids of the maximum likelihood estimator.
- [587] arXiv:2312.05984 (replaced) [pdf, html, other]
-
Title: Accurate Differential Operators for Hybrid Neural FieldsComments: Accepted in CVPR 2025. Project page is available at this https URLSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Graphics (cs.GR); Machine Learning (cs.LG)
Neural fields have become widely used in various fields, from shape representation to neural rendering, and for solving partial differential equations (PDEs). With the advent of hybrid neural field representations like Instant NGP that leverage small MLPs and explicit representations, these models train quickly and can fit large scenes. Yet in many applications like rendering and simulation, hybrid neural fields can cause noticeable and unreasonable artifacts. This is because they do not yield accurate spatial derivatives needed for these downstream applications. In this work, we propose two ways to circumvent these challenges. Our first approach is a post hoc operator that uses local polynomial fitting to obtain more accurate derivatives from pre-trained hybrid neural fields. Additionally, we also propose a self-supervised fine-tuning approach that refines the hybrid neural field to yield accurate derivatives directly while preserving the initial signal. We show applications of our method to rendering, collision simulation, and solving PDEs. We observe that using our approach yields more accurate derivatives, reducing artifacts and leading to more accurate simulations in downstream applications.
- [588] arXiv:2312.06562 (replaced) [pdf, html, other]
-
Title: On Meta-PromptingSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Category Theory (math.CT)
Modern large language models (LLMs) are capable of interpreting input strings as instructions, or prompts, and carry out tasks based on them. Unlike traditional learners, LLMs cannot use back-propagation to obtain feedback, and condition their output in situ in a phenomenon known as in-context learning (ICL). Many approaches to prompting and pre-training these models involve the automated generation of these prompts, also known as meta-prompting, or prompting to obtain prompts. However, they do not formally describe the properties and behavior of the LLMs themselves. We propose a theoretical framework based on category theory to generalize and describe ICL and LLM behavior when interacting with users. Our framework allows us to obtain formal results around task agnosticity and equivalence of various meta-prompting approaches. Using our framework and experimental results we argue that meta-prompting is more effective than basic prompting at generating desirable outputs.
- [589] arXiv:2401.16092 (replaced) [pdf, html, other]
-
Title: Multilingual Text-to-Image Generation Magnifies Gender Stereotypes and Prompt Engineering May Not Help YouFelix Friedrich, Katharina Hämmerl, Patrick Schramowski, Manuel Brack, Jindrich Libovicky, Kristian Kersting, Alexander FraserSubjects: Computation and Language (cs.CL); Computers and Society (cs.CY); Machine Learning (cs.LG)
Text-to-image generation models have recently achieved astonishing results in image quality, flexibility, and text alignment, and are consequently employed in a fast-growing number of applications. Through improvements in multilingual abilities, a larger community now has access to this technology. However, our results show that multilingual models suffer from significant gender biases just as monolingual models do. Furthermore, the natural expectation that multilingual models will provide similar results across languages does not hold up. Instead, there are important differences between languages. We propose a novel benchmark, MAGBIG, intended to foster research on gender bias in multilingual models. We use MAGBIG to investigate the effect of multilingualism on gender bias in T2I models. To this end, we construct multilingual prompts requesting portraits of people with a certain occupation or trait. Our results show that not only do models exhibit strong gender biases but they also behave differently across languages. Furthermore, we investigate prompt engineering strategies, such as indirect, neutral formulations, to mitigate these biases. Unfortunately, these approaches have limited success and result in worse text-to-image alignment. Consequently, we call for more research into diverse representations across languages in image generators, as well as into steerability to address biased model behavior.
- [590] arXiv:2402.07723 (replaced) [pdf, html, other]
-
Title: Generalization Bounds for Heavy-Tailed SDEs through the Fractional Fokker-Planck EquationJournal-ref: Proceedings of the 41st International Conference on Machine Learning, PMLR 235:12087-12137, 2024Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
Understanding the generalization properties of heavy-tailed stochastic optimization algorithms has attracted increasing attention over the past years. While illuminating interesting aspects of stochastic optimizers by using heavy-tailed stochastic differential equations as proxies, prior works either provided expected generalization bounds, or introduced non-computable information theoretic terms. Addressing these drawbacks, in this work, we prove high-probability generalization bounds for heavy-tailed SDEs which do not contain any nontrivial information theoretic terms. To achieve this goal, we develop new proof techniques based on estimating the entropy flows associated with the so-called fractional Fokker-Planck equation (a partial differential equation that governs the evolution of the distribution of the corresponding heavy-tailed SDE). In addition to obtaining high-probability bounds, we show that our bounds have a better dependence on the dimension of parameters as compared to prior art. Our results further identify a phase transition phenomenon, which suggests that heavy tails can be either beneficial or harmful depending on the problem structure. We support our theory with experiments conducted in a variety of settings.
- [591] arXiv:2402.08151 (replaced) [pdf, html, other]
-
Title: Perturbative partial moment matching and gradient-flow adaptive importance sampling transformations for Bayesian leave one out cross-validationComments: SubmittedSubjects: Methodology (stat.ME); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Spectral Theory (math.SP); Statistics Theory (math.ST)
Importance sampling (IS) allows one to approximate leave one out (LOO) cross-validation for a Bayesian model, without refitting, by inverting the Bayesian update equation to subtract a given data point from a model posterior. For each data point, one computes expectations under the corresponding LOO posterior by weighted averaging over the full data posterior. This task sometimes requires weight stabilization in the form of adapting the posterior distribution via transformation. So long as one is successful in finding a suitable transformation, one avoids refitting. To this end, we motivate the use of bijective perturbative transformations of the form $T(\boldsymbol{\theta})=\boldsymbol{\theta} + h Q(\boldsymbol{\theta}),$ for $0<h\ll 1,$ and introduce two classes of such transformations: 1) partial moment matching and 2) gradient flow evolution. The former extends prior literature on moment-matching under the recognition that adaptation for LOO is a small perturbation on the full data posterior. The latter class of methods define transformations based on relaxing various statistical objectives: in our case the variance of the IS estimator and the KL divergence between the transformed distribution and the statistics of the LOO fold. Being model-specific, the gradient flow transformations require evaluating Jacobian determinants. While these quantities are generally readily available through auto-differentiation, we derive closed-form expressions in the case of logistic regression and shallow ReLU activated neural networks. We tested the methodology on an $n\ll p$ dataset that is known to produce unstable LOO IS weights.
- [592] arXiv:2402.10252 (replaced) [pdf, html, other]
-
Title: Online Control of Linear Systems under Unbounded NoiseComments: 41 pagesSubjects: Systems and Control (eess.SY); Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
This paper investigates the problem of controlling a linear system under possibly unbounded stochastic noise with unknown convex cost functions, known as an online control problem. In contrast to the existing work, which assumes the boundedness of noise, we show that an $ \tilde{O}(\sqrt{T}) $ high-probability regret can be achieved under unbounded noise, where $ T $ denotes the time horizon. Notably, the noise is only required to have a finite fourth moment. Moreover, when the costs are strongly convex and the noise is sub-Gaussian, we establish an $ O({\rm poly} (\log T)) $ regret bound.
- [593] arXiv:2403.11981 (replaced) [pdf, html, other]
-
Title: Certified Robustness to Clean-Label Poisoning Using Diffusion DenoisingSubjects: Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
We present a certified defense to clean-label poisoning attacks under $\ell_2$-norm. These attacks work by injecting a small number of poisoning samples (e.g., 1%) that contain bounded adversarial perturbations into the training data to induce a targeted misclassification of a test-time input. Inspired by the adversarial robustness achieved by $randomized$ $smoothing$, we show how an off-the-shelf diffusion denoising model can sanitize the tampered training data. We extensively test our defense against seven clean-label poisoning attacks in both $\ell_2$ and $\ell_{\infty}$-norms and reduce their attack success to 0-16% with only a negligible drop in the test accuracy. We compare our defense with existing countermeasures against clean-label poisoning, showing that the defense reduces the attack success the most and offers the best model utility. Our results highlight the need for future work on developing stronger clean-label attacks and using our certified yet practical defense as a strong baseline to evaluate these attacks.
- [594] arXiv:2404.10304 (replaced) [pdf, html, other]
-
Title: LLM-Powered Test Case Generation for Detecting Bugs in Plausible ProgramsKaibo Liu, Zhenpeng Chen, Yiyang Liu, Jie M. Zhang, Mark Harman, Yudong Han, Yun Ma, Yihong Dong, Ge Li, Gang HuangComments: Accepted by the 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025) Main TrackSubjects: Software Engineering (cs.SE); Machine Learning (cs.LG)
Detecting tricky bugs in plausible programs, those that pass existing test suites yet still contain bugs, remains a significant challenge in software testing. To address this problem, we propose TrickCatcher, an LLM-powered approach to generating test cases for uncovering bugs in plausible programs. TrickCatcher operates in three stages: First, it uses an LLM to generate program variants based on the program under test (PUT) and its specification. Second, it employs an LLM to construct an input generator from the specification for producing test inputs. Finally, these inputs are executed on both the PUT and its program variants to detect inconsistencies in their outputs. We evaluate TrickCatcher on two datasets, TrickyBugs and EvalPlus, which include 366 human-written and 151 AI-generated plausible programs with tricky bugs. TrickCatcher achieves recall, precision, and F1 scores that are 1.80x, 2.65x, and 1.66x those of the state-of-the-art baselines, respectively. Code and data used are available at this https URL.
- [595] arXiv:2405.03205 (replaced) [pdf, html, other]
-
Title: Anchored Answers: Unravelling Positional Bias in GPT-2's Multiple-Choice QuestionsComments: ACL 2025 FindingsSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Large Language Models (LLMs), such as the GPT-4 and LLaMA families, have demonstrated considerable success across diverse tasks, including multiple-choice questions (MCQs). However, these models exhibit a positional bias, particularly an even worse anchored bias in the GPT-2 family, where they consistently favour the first choice 'A' in MCQs during inference. This anchored bias challenges the integrity of GPT-2's decision-making process, as it skews performance based on the position rather than the content of the choices in MCQs. In this study, we utilise the mechanistic interpretability approach to identify the internal modules within GPT-2 models responsible for this bias. We focus on the Multi-Layer Perceptron (MLP) layers and attention heads, using the "logit lens" method to trace and modify the specific value vectors that contribute to the bias. By updating these vectors within MLP and recalibrating attention patterns to neutralise the preference for the first choice 'A', we effectively mitigate the anchored bias. Our interventions not only mitigate the bias but also improve the overall MCQ prediction accuracy for the GPT-2 family across various datasets. This work represents the first comprehensive mechanistic analysis of anchored bias from the failing cases in MCQs within the GPT-2 models, introducing targeted, minimal-intervention strategies that significantly enhance GPT2 model robustness and accuracy in MCQs. Our code is available at this https URL.
- [596] arXiv:2405.16837 (replaced) [pdf, other]
-
Title: Enhancing Accuracy in Generative Models via Knowledge TransferComments: 57pages, 5 figuresSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
This paper investigates the accuracy of generative models and the impact of knowledge transfer on their generation precision. Specifically, we examine a generative model for a target task, fine-tuned using a pre-trained model from a source task. Building on the "Shared Embedding" concept, which bridges the source and target tasks, we introduce a novel framework for transfer learning under distribution metrics such as the Kullback-Leibler divergence. This framework underscores the importance of leveraging inherent similarities between diverse tasks despite their distinct data distributions. Our theory suggests that the shared structures can augment the generation accuracy for a target task, reliant on the capability of a source model to identify shared structures and effective knowledge transfer from source to target learning. To demonstrate the practical utility of this framework, we explore the theoretical implications for two specific generative models: diffusion and normalizing flows. The results show enhanced performance in both models over their non-transfer counterparts, indicating advancements for diffusion models and providing fresh insights into normalizing flows in transfer and non-transfer settings. These results highlight the significant contribution of knowledge transfer in boosting the generation capabilities of these models.
- [597] arXiv:2406.02394 (replaced) [pdf, other]
-
Title: Pattern Recognition or Medical Knowledge? The Problem with Multiple-Choice Questions in MedicineComments: ACL 2025 mainSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Large Language Models (LLMs) such as ChatGPT demonstrate significant potential in the medical domain and are often evaluated using multiple-choice questions (MCQs) modeled on exams like the USMLE. However, such benchmarks may overestimate true clinical understanding by rewarding pattern recognition and test-taking heuristics. To investigate this, we created a fictional medical benchmark centered on an imaginary organ, the Glianorex, allowing us to separate memorized knowledge from reasoning ability. We generated textbooks and MCQs in English and French using leading LLMs, then evaluated proprietary, open-source, and domain-specific models in a zero-shot setting. Despite the fictional content, models achieved an average score of 64%, while physicians scored only 27%. Fine-tuned medical models outperformed base models in English but not in French. Ablation and interpretability analyses revealed that models frequently relied on shallow cues, test-taking strategies, and hallucinated reasoning to identify the correct choice. These results suggest that standard MCQ-based evaluations may not effectively measure clinical reasoning and highlight the need for more robust, clinically meaningful assessment methods for LLMs.
- [598] arXiv:2406.05694 (replaced) [pdf, html, other]
-
Title: A Low Rank Neural Representation of Entropy SolutionsComments: 42 pages, 9 figuresSubjects: Numerical Analysis (math.NA); Machine Learning (cs.LG)
We construct a new representation of entropy solutions to nonlinear scalar conservation laws with a smooth convex flux function in a single spatial dimension. The representation is a generalization of the method of characteristics and posseses a compositional form. While it is a nonlinear representation, the embedded dynamics of the solution in the time variable is linear. This representation is then discretized as a manifold of implicit neural representations where the feedforward neural network architecture has a low rank structure. Finally, we show that the low rank neural representation with a fixed number of layers and a small number of coefficients can approximate any entropy solution regardless of the complexity of the shock topology, while retaining the linearity of the embedded dynamics.
- [599] arXiv:2406.11753 (replaced) [pdf, other]
-
Title: A Semantic-Aware Layer-Freezing Approach to Computation-Efficient Fine-Tuning of Language ModelsComments: accepted by ACL 2025, the camera-ready versionSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Finetuning language models (LMs) is crucial for adapting the models to downstream data and tasks. However, full finetuning is usually costly. Existing work, such as parameter-efficient finetuning (PEFT), often focuses on \textit{how to finetune} but neglects the issue of \textit{where to finetune}. As a pioneering work on reducing the cost of backpropagation (at the layer level) by answering where to finetune, we conduct a semantic analysis of the LM inference process. We first propose using transition traces of the latent representation to compute deviations (or loss). Then, using a derived formula of scaling law, we estimate the gain of each layer in reducing deviation (or loss). Further, we narrow down the scope for finetuning, and also, study the cost-benefit balance of LM finetuning. We perform extensive experiments across well-known LMs and datasets. The results show that our approach is effective and efficient, and outperforms the existing baselines. Our approach is orthogonal to other techniques for improving finetuning efficiency, such as PEFT methods, offering practical values on LM finetuning.
- [600] arXiv:2406.13945 (replaced) [pdf, html, other]
-
Title: CityBench: Evaluating the Capabilities of Large Language Models for Urban TasksJie Feng, Jun Zhang, Tianhui Liu, Xin Zhang, Tianjian Ouyang, Junbo Yan, Yuwei Du, Siqi Guo, Yong LiComments: Accepted by KDD 2025 D&B Track, this https URLSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
As large language models (LLMs) continue to advance and gain widespread use, establishing systematic and reliable evaluation methodologies for LLMs and vision-language models (VLMs) has become essential to ensure their real-world effectiveness and reliability. There have been some early explorations about the usability of LLMs for limited urban tasks, but a systematic and scalable evaluation benchmark is still lacking. The challenge in constructing a systematic evaluation benchmark for urban research lies in the diversity of urban data, the complexity of application scenarios and the highly dynamic nature of the urban environment. In this paper, we design \textit{CityBench}, an interactive simulator based evaluation platform, as the first systematic benchmark for evaluating the capabilities of LLMs for diverse tasks in urban research. First, we build \textit{CityData} to integrate the diverse urban data and \textit{CitySimu} to simulate fine-grained urban dynamics. Based on \textit{CityData} and \textit{CitySimu}, we design 8 representative urban tasks in 2 categories of perception-understanding and decision-making as the \textit{CityBench}. With extensive results from 30 well-known LLMs and VLMs in 13 cities around the world, we find that advanced LLMs and VLMs can achieve competitive performance in diverse urban tasks requiring commonsense and semantic understanding abilities, e.g., understanding the human dynamics and semantic inference of urban images. Meanwhile, they fail to solve the challenging urban tasks requiring professional knowledge and high-level numerical abilities, e.g., geospatial prediction and traffic control task.
- [601] arXiv:2406.13948 (replaced) [pdf, html, other]
-
Title: CityGPT: Empowering Urban Spatial Cognition of Large Language ModelsComments: Accepted by KDD 2025 Research Track, this https URLSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Large language models(LLMs), with their powerful language generation and reasoning capabilities, have already achieved notable success in many domains, e.g., math and code generation. However, they often fall short when tackling real-life geospatial tasks within urban environments. This limitation stems from a lack of physical world knowledge and relevant data during training. To address this gap, we propose \textit{CityGPT}, a systematic framework designed to enhance LLMs' understanding of urban space and improve their ability to solve the related urban tasks by integrating a city-scale `world model' into the model. Firstly, we construct a diverse instruction tuning dataset, \textit{CityInstruction}, for injecting urban knowledge into LLMs and effectively boosting their spatial reasoning capabilities. Using a combination of \textit{CityInstruction} and open source general instruction data, we introduce a novel and easy-to-use self-weighted fine-tuning method (\textit{SWFT}) to train various LLMs (including ChatGLM3-6B, Llama3-8B, and Qwen2.5-7B) to enhance their urban spatial capabilities without compromising, or even improving, their general abilities. Finally, to validate the effectiveness of our proposed framework, we develop a comprehensive text-based spatial benchmark \textit{CityEval} for evaluating the performance of LLMs across a wide range of urban scenarios and geospatial tasks. Extensive evaluation results demonstrate that smaller LLMs trained with \textit{CityInstruction} by \textit{SWFT} method can achieve performance that is competitive with, and in some cases superior to, proprietary LLMs when assessed using \textit{CityEval}.
- [602] arXiv:2407.17491 (replaced) [pdf, html, other]
-
Title: Robust Adaptation of Foundation Models with Black-Box Visual PromptingComments: Extended work from the CVPR'23 paper: arXiv:2303.14773; This paper has been submitted to IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI) for possible publicationSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
With a surge of large-scale pre-trained models, parameter-efficient transfer learning (PETL) of large models has garnered significant attention. While promising, they commonly rely on two optimistic assumptions: 1) full access to the parameters of a PTM, and 2) sufficient memory capacity to cache all intermediate activations for gradient computation. However, in most real-world applications, PTMs serve as black-box APIs or proprietary software without full parameter accessibility. Besides, it is hard to meet a large memory requirement for modern PTMs. This work proposes black-box visual prompting (BlackVIP), which efficiently adapts the PTMs without knowledge of their architectures or parameters. BlackVIP has two components: 1) Coordinator and 2) simultaneous perturbation stochastic approximation with gradient correction (SPSA-GC). The Coordinator designs input-dependent visual prompts, which allow the target PTM to adapt in the wild. SPSA-GC efficiently estimates the gradient of PTM to update Coordinator. Besides, we introduce a variant, BlackVIP-SE, which significantly reduces the runtime and computational cost of BlackVIP. Extensive experiments on 19 datasets demonstrate that BlackVIPs enable robust adaptation to diverse domains and tasks with minimal memory requirements. We further provide a theoretical analysis on the generalization of visual prompting methods by presenting their connection to the certified robustness of randomized smoothing, and presenting an empirical support for improved robustness.
- [603] arXiv:2408.04607 (replaced) [pdf, html, other]
-
Title: Risk and cross validation in ridge regression with correlated samplesComments: 44 pages, 19 figures. v4: ICML 2025 camera-readySubjects: Machine Learning (stat.ML); Disordered Systems and Neural Networks (cond-mat.dis-nn); Machine Learning (cs.LG)
Recent years have seen substantial advances in our understanding of high-dimensional ridge regression, but existing theories assume that training examples are independent. By leveraging techniques from random matrix theory and free probability, we provide sharp asymptotics for the in- and out-of-sample risks of ridge regression when the data points have arbitrary correlations. We demonstrate that in this setting, the generalized cross validation estimator (GCV) fails to correctly predict the out-of-sample risk. However, in the case where the noise residuals have the same correlations as the data points, one can modify the GCV to yield an efficiently-computable unbiased estimator that concentrates in the high-dimensional limit, which we dub CorrGCV. We further extend our asymptotic analysis to the case where the test point has nontrivial correlations with the training set, a setting often encountered in time series forecasting. Assuming knowledge of the correlation structure of the time series, this again yields an extension of the GCV estimator, and sharply characterizes the degree to which such test points yield an overly optimistic prediction of long-time risk. We validate the predictions of our theory across a variety of high dimensional data.
- [604] arXiv:2408.09181 (replaced) [pdf, html, other]
-
Title: PADetBench: Towards Benchmarking Physical Attacks against Object DetectionSubjects: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
Physical attacks against object detection have gained increasing attention due to their significant practical implications. However, conducting physical experiments is extremely time-consuming and labor-intensive. Moreover, physical dynamics and cross-domain transformation are challenging to strictly regulate in the real world, leading to unaligned evaluation and comparison, severely hindering the development of physically robust models. To accommodate these challenges, we explore utilizing realistic simulation to thoroughly and rigorously benchmark physical attacks with fairness under controlled physical dynamics and cross-domain transformation. This resolves the problem of capturing identical adversarial images that cannot be achieved in the real world. Our benchmark includes 20 physical attack methods, 48 object detectors, comprehensive physical dynamics, and evaluation metrics. We also provide end-to-end pipelines for dataset generation, detection, evaluation, and further analysis. In addition, we perform 8064 groups of evaluation based on our benchmark, which includes both overall evaluation and further detailed ablation studies for controlled physical dynamics. Through these experiments, we provide in-depth analyses of physical attack performance and physical adversarial robustness, draw valuable observations, and discuss potential directions for future research.
Codebase: this https URL - [605] arXiv:2408.16028 (replaced) [pdf, html, other]
-
Title: ANVIL: Anomaly-based Vulnerability Identification without Labelled Training DataSubjects: Cryptography and Security (cs.CR); Machine Learning (cs.LG); Software Engineering (cs.SE)
Supervised-learning-based vulnerability detectors often fall short due to limited labelled training data. In contrast, Large Language Models (LLMs) like GPT-4 are trained on vast unlabelled code corpora, yet perform only marginally better than coin flips when directly prompted to detect vulnerabilities. In this paper, we reframe vulnerability detection as anomaly detection, based on the premise that vulnerable code is rare and thus anomalous relative to patterns learned by LLMs. We introduce ANVIL, which performs a masked code reconstruction task: the LLM reconstructs a masked line of code, and deviations from the original are scored as anomalies. We propose a hybrid anomaly score that combines exact match, cross-entropy loss, prediction confidence, and structural complexity. We evaluate our approach across multiple LLM families, scoring methods, and context sizes, and against vulnerabilities after the LLM's training cut-off. On the PrimeVul dataset, ANVIL outperforms state-of-the-art supervised detectors-LineVul, LineVD, and LLMAO-achieving up to 2x higher Top-3 accuracy, 75% better Normalized MFR, and a significant improvement on ROC-AUC. Finally, by integrating ANVIL with fuzzers, we uncover two previously unknown vulnerabilities, demonstrating the practical utility of anomaly-guided detection.
- [606] arXiv:2408.17151 (replaced) [pdf, html, other]
-
Title: Investigating Privacy Leakage in Dimensionality Reduction Methods via Reconstruction AttackComments: Journal of Information Security and Applications, 2025, vol. 92, pp. 104102Subjects: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
This study investigates privacy leakage in dimensionality reduction methods through a novel machine learning-based reconstruction attack. Employing an informed adversary threat model, we develop a neural network capable of reconstructing high-dimensional data from low-dimensional embeddings.
We evaluate six popular dimensionality reduction techniques: principal component analysis (PCA), sparse random projection (SRP), multidimensional scaling (MDS), Isomap, $t$-distributed stochastic neighbor embedding ($t$-SNE), and uniform manifold approximation and projection (UMAP). Using both MNIST and NIH Chest X-ray datasets, we perform a qualitative analysis to identify key factors affecting reconstruction quality. Furthermore, we assess the effectiveness of an additive noise mechanism in mitigating these reconstruction attacks. Our experimental results on both datasets reveal that the attack is effective against deterministic methods (PCA and Isomap). but ineffective against methods that employ random initialization (SRP, MDS, $t$-SNE and UMAP). The experimental results also show that, for PCA and Isomap, our reconstruction network produces higher quality outputs compared to a previously proposed network.
We also study the effect of additive noise mechanism to prevent the reconstruction attack. Our experiment shows that, when adding the images with large noises before performing PCA or Isomap, the attack produced severely distorted reconstructions. In contrast, for the other four methods, the reconstructions still show some recognizable features, though they bear little resemblance to the original images. The code is available at this https URL - [607] arXiv:2409.03219 (replaced) [pdf, html, other]
-
Title: Content Moderation by LLM: From Accuracy to LegitimacySubjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
One trending application of LLM (large language model) is to use it for content moderation in online platforms. Most current studies on this application have focused on the metric of accuracy -- the extent to which LLMs make correct decisions about content. This article argues that accuracy is insufficient and misleading because it fails to grasp the distinction between easy cases and hard cases, as well as the inevitable trade-offs in achieving higher accuracy. Closer examination reveals that content moderation is a constitutive part of platform governance, the key of which is to gain and enhance legitimacy. Instead of making moderation decisions correct, the chief goal of LLMs is to make them legitimate. In this regard, this article proposes a paradigm shift from the single benchmark of accuracy towards a legitimacy-based framework for evaluating the performance of LLM moderators. The framework suggests that for easy cases, the key is to ensure accuracy, speed, and transparency, while for hard cases, what matters is reasoned justification and user participation. Examined under this framework, LLMs' real potential in moderation is not accuracy improvement. Rather, LLMs can better contribute in four other aspects: to conduct screening of hard cases from easy cases, to provide quality explanations for moderation decisions, to assist human reviewers in getting more contextual information, and to facilitate user participation in a more interactive way. To realize these contributions, this article proposes a workflow for incorporating LLMs into the content moderation system. Using normative theories from law and social sciences to critically assess the new technological application, this article seeks to redefine LLMs' role in content moderation and redirect relevant research in this field.
- [608] arXiv:2409.03685 (replaced) [pdf, html, other]
-
Title: View-Invariant Policy Learning via Zero-Shot Novel View SynthesisComments: Accepted to CoRL 2024Subjects: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Large-scale visuomotor policy learning is a promising approach toward developing generalizable manipulation systems. Yet, policies that can be deployed on diverse embodiments, environments, and observational modalities remain elusive. In this work, we investigate how knowledge from large-scale visual data of the world may be used to address one axis of variation for generalizable manipulation: observational viewpoint. Specifically, we study single-image novel view synthesis models, which learn 3D-aware scene-level priors by rendering images of the same scene from alternate camera viewpoints given a single input image. For practical application to diverse robotic data, these models must operate zero-shot, performing view synthesis on unseen tasks and environments. We empirically analyze view synthesis models within a simple data-augmentation scheme that we call View Synthesis Augmentation (VISTA) to understand their capabilities for learning viewpoint-invariant policies from single-viewpoint demonstration data. Upon evaluating the robustness of policies trained with our method to out-of-distribution camera viewpoints, we find that they outperform baselines in both simulated and real-world manipulation tasks. Videos and additional visualizations are available at this https URL.
- [609] arXiv:2409.04459 (replaced) [pdf, html, other]
-
Title: WET: Overcoming Paraphrasing Vulnerabilities in Embeddings-as-a-Service with Linear Transformation WatermarksComments: Accepted to ACL 2025 (Main Proceedings)Subjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL); Machine Learning (cs.LG)
Embeddings-as-a-Service (EaaS) is a service offered by large language model (LLM) developers to supply embeddings generated by LLMs. Previous research suggests that EaaS is prone to imitation attacks -- attacks that clone the underlying EaaS model by training another model on the queried embeddings. As a result, EaaS watermarks are introduced to protect the intellectual property of EaaS providers. In this paper, we first show that existing EaaS watermarks can be removed by paraphrasing when attackers clone the model. Subsequently, we propose a novel watermarking technique that involves linearly transforming the embeddings, and show that it is empirically and theoretically robust against paraphrasing.
- [610] arXiv:2409.05806 (replaced) [pdf, html, other]
-
Title: CKnowEdit: A New Chinese Knowledge Editing Dataset for Linguistics, Facts, and Logic Error Correction in LLMsComments: ACL 2025; project website is available at this https URL code and dataset are available at this https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Machine Learning (cs.LG)
Chinese, as a linguistic system rich in depth and complexity, is characterized by distinctive elements such as ancient poetry, proverbs, idioms, and other cultural constructs. However, current Large Language Models (LLMs) face limitations in these specialized domains, highlighting the need for the development of comprehensive datasets that can assess, continuously update, and progressively improve these culturally-grounded linguistic competencies through targeted training optimizations. To address this gap, we introduce CKnowEdit, the first-ever Chinese knowledge editing dataset designed to correct linguistic, factual, and logical errors in LLMs. We collect seven types of knowledge from a wide range of sources, including classical texts, idioms, and content from Baidu Tieba Ruozhiba, taking into account the unique polyphony, antithesis, and logical structures inherent in the Chinese language. By analyzing this dataset, we highlight the challenges current LLMs face in mastering Chinese. Furthermore, our evaluation of state-of-the-art knowledge editing techniques reveals opportunities to advance the correction of Chinese knowledge. Code and dataset are available at this https URL.
- [611] arXiv:2409.10289 (replaced) [pdf, html, other]
-
Title: ReflectDiffu:Reflect between Emotion-intent Contagion and Mimicry for Empathetic Response Generation via a RL-Diffusion FrameworkComments: Accepted by ACL 2025 Main ConferenceSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Empathetic response generation necessitates the integration of emotional and intentional dynamics to foster meaningful interactions. Existing research either neglects the intricate interplay between emotion and intent, leading to suboptimal controllability of empathy, or resorts to large language models (LLMs), which incur significant computational overhead. In this paper, we introduce ReflectDiffu, a lightweight and comprehensive framework for empathetic response generation. This framework incorporates emotion contagion to augment emotional expressiveness and employs an emotion-reasoning mask to pinpoint critical emotional elements. Additionally, it integrates intent mimicry within reinforcement learning for refinement during diffusion. By harnessing an intent twice reflect mechanism of Exploring-Sampling-Correcting, ReflectDiffu adeptly translates emotional decision-making into precise intent actions, thereby addressing empathetic response misalignments stemming from emotional misrecognition. Through reflection, the framework maps emotional states to intents, markedly enhancing both response empathy and flexibility. Comprehensive experiments reveal that ReflectDiffu outperforms existing models regarding relevance, controllability, and informativeness, achieving state-of-the-art results in both automatic and human evaluations.
- [612] arXiv:2409.19458 (replaced) [pdf, html, other]
-
Title: Scalable Fine-tuning from Multiple Data Sources: A First-Order Approximation ApproachComments: 17 pages. Appeared in Findings of EMNLP'24Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
We study the problem of fine-tuning a language model (LM) for a target task by optimally using the information from $n$ auxiliary tasks. This problem has broad applications in NLP, such as targeted instruction tuning and data selection in chain-of-thought fine-tuning. The key challenge of this problem is that not all auxiliary tasks are beneficial in improving the performance of the target task. Thus, selecting the right subset of auxiliary tasks is crucial. Conventional subset selection methods, such as forward and backward stepwise selection, are unsuitable for LM fine-tuning because they require repeated training on subsets of auxiliary tasks. This paper introduces a new algorithm for estimating model fine-tuning performance without requiring repeated training. Our algorithm first performs multitask training using data from all tasks to obtain a meta initialization. Then, we approximate the model fine-tuning loss of a subset using functional values and gradients from the meta initialization. Empirically, we find that this gradient-based approximation holds with remarkable accuracy for twelve transformer-based LMs. Thus, we can now estimate fine-tuning performances on CPUs within a few seconds. Finally, we fine-tune the pretrained base model once on the selected subset of tasks. We conduct extensive experiments to validate this approach, delivering a speedup of $30\times$ over conventional subset selection while incurring only $1\%$ error of the true fine-tuning performances. In downstream evaluations involving both instruction tuning and chain-of-thought fine-tuning, this loss-based selection approach improves over prior gradient or representation similarity-based methods for subset selection by up to $3.8\%$.
- [613] arXiv:2410.03026 (replaced) [pdf, html, other]
-
Title: Estimating Privacy Leakage of Augmented Contextual Knowledge in Language ModelsSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
Language models (LMs) rely on their parametric knowledge augmented with relevant contextual knowledge for certain tasks, such as question answering. However, the contextual knowledge can contain private information that may be leaked when answering queries, and estimating this privacy leakage is not well understood. A straightforward approach of directly comparing an LM's output to the contexts can overestimate the privacy risk, since the LM's parametric knowledge might already contain the augmented contextual knowledge. To this end, we introduce $\emph{context influence}$, a metric that builds on differential privacy, a widely-adopted privacy notion, to estimate the privacy leakage of contextual knowledge during decoding. Our approach effectively measures how each subset of the context influences an LM's response while separating the specific parametric knowledge of the LM. Using our context influence metric, we demonstrate that context privacy leakage occurs when contextual knowledge is out of distribution with respect to parametric knowledge. Moreover, we experimentally demonstrate how context influence properly attributes the privacy leakage to augmented contexts, and we evaluate how factors-- such as model size, context size, generation position, etc.-- affect context privacy leakage. The practical implications of our results will inform practitioners of the privacy risk associated with augmented contextual knowledge.
- [614] arXiv:2410.07176 (replaced) [pdf, html, other]
-
Title: Astute RAG: Overcoming Imperfect Retrieval Augmentation and Knowledge Conflicts for Large Language ModelsComments: ACL 2025 main conferenceSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Retrieval augmented generation (RAG), while effectively integrating external knowledge to address the inherent limitations of large language models (LLMs), can be hindered by imperfect retrieval that contain irrelevant, misleading, or even malicious information. Previous studies have rarely connected the behavior of RAG through joint analysis, particularly regarding error propagation coming from imperfect retrieval and potential conflicts between LLMs' internal knowledge and external sources. Through comprehensive and controlled analyses under realistic conditions, we find that imperfect retrieval augmentation is inevitable, common, and harmful. We identify the knowledge conflicts between LLM-internal and external knowledge from retrieval as a bottleneck to overcome imperfect retrieval in the post-retrieval stage of RAG. To address this, we propose Astute RAG, a novel RAG approach designed to be resilient to imperfect retrieval augmentation. It adaptively elicits essential information from LLMs' internal knowledge, iteratively consolidates internal and external knowledge with source-awareness, and finalizes the answer according to information reliability. Our experiments with Gemini and Claude demonstrate the superior performance of Astute RAG compared to previous robustness-enhanced RAG approaches. Specifically, Astute RAG is the only RAG method that achieves performance comparable to or even surpassing conventional use of LLMs under the worst-case scenario. Further analysis reveals the effectiveness of Astute RAG in resolving knowledge conflicts, thereby improving the trustworthiness of RAG.
- [615] arXiv:2410.08934 (replaced) [pdf, html, other]
-
Title: Understanding the Statistical Accuracy-Communication Trade-off in Personalized Federated Learning with Minimax GuaranteesComments: Published in Proceedings of the 42st International Conference on Machine Learning (ICML 2025)Subjects: Machine Learning (stat.ML); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG); Statistics Theory (math.ST); Computation (stat.CO)
Personalized federated learning (PFL) offers a flexible framework for aggregating information across distributed clients with heterogeneous data. This work considers a personalized federated learning setting that simultaneously learns global and local models. While purely local training has no communication cost, collaborative learning among the clients can leverage shared knowledge to improve statistical accuracy, presenting an accuracy-communication trade-off in personalized federated learning. However, the theoretical analysis of how personalization quantitatively influences sample and algorithmic efficiency and their inherent trade-off is largely unexplored. This paper makes a contribution towards filling this gap, by providing a quantitative characterization of the personalization degree on the tradeoff. The results further offers theoretical insights for choosing the personalization degree. As a side contribution, we establish the minimax optimality in terms of statistical accuracy for a widely studied PFL formulation. The theoretical result is validated on both synthetic and real-world datasets and its generalizability is verified in a non-convex setting.
- [616] arXiv:2410.12201 (replaced) [pdf, html, other]
-
Title: Data-light Uncertainty Set Merging with Admissibility: Synthetics, Aggregation, and Test InversionSubjects: Methodology (stat.ME); Machine Learning (cs.LG)
This article introduces a Synthetics, Aggregation, and Test inversion (SAT) approach for merging diverse and potentially dependent uncertainty sets into a single unified set. The procedure is data-light, relying only on initial sets and control levels, and it adapts to any user-specified initial uncertainty sets, accommodating potentially varying coverage levels. SAT is motivated by the challenge of integrating uncertainty sets when only the initial sets and their control levels are available - for example, when merging confidence sets from distributed sites under communication constraints or combining conformal prediction sets generated by different algorithms or data splits. To address this, SAT constructs and aggregates novel synthetic test statistics, and then derive merged sets through test inversion. Our method leverages the duality between set estimation and hypothesis testing, ensuring reliable coverage in dependent scenarios. A key theoretical contribution is a rigorous analysis of SAT's properties, including a proof of its admissibility in the context of deterministic set merging. Both theoretical analyses and empirical results confirm the method's finite-sample coverage validity and desirable set sizes.
- [617] arXiv:2410.12613 (replaced) [pdf, html, other]
-
Title: Exploring Model Kinship for Merging Large Language ModelsComments: Ongoing workSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
Model merging has become one of the key technologies for enhancing the capabilities and efficiency of Large Language Models (LLMs). However, our understanding of the expected performance gains and principles when merging any two models remains limited. In this work, we introduce model kinship, the degree of similarity or relatedness between LLMs, analogous to biological evolution. With comprehensive empirical analysis, we find that there is a certain relationship between model kinship and the performance gains after model merging, which can help guide our selection of candidate models. Inspired by this, we propose a new model merging strategy: Top-k Greedy Merging with Model Kinship, which can yield better performance on benchmark datasets. Specifically, we discover that using model kinship as a criterion can assist us in continuously performing model merging, alleviating the degradation (local optima) in model evolution, whereas model kinship can serve as a guide to escape these traps. Code is available at this https URL.
- [618] arXiv:2410.13098 (replaced) [pdf, html, other]
-
Title: A Little Human Data Goes A Long WayComments: ACL 2025Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Faced with an expensive human annotation process, creators of NLP systems increasingly turn to synthetic data generation. While this method shows promise, the extent to which synthetic data can replace human annotation is poorly understood. We investigate the use of synthetic data in Fact Verification (FV) and Question Answering (QA) by studying the effects of incrementally replacing human generated data with synthetic points on eight diverse datasets. Strikingly, replacing up to 90% of the training data only marginally decreases performance, but replacing the final 10% leads to severe declines. We find that models trained on purely synthetic data can be reliably improved by including as few as 125 human generated data points. We show that matching the performance gain of just a little additional human data (only 200 points) requires an order of magnitude more synthetic data and estimate price ratios at which human annotation would be a more cost-effective solution. Our results suggest that even when human annotation at scale is infeasible, there is great value to having a small proportion of the dataset being human generated.
- [619] arXiv:2410.14673 (replaced) [pdf, html, other]
-
Title: Self-supervised contrastive learning performs non-linear system identificationComments: Published as a conference paper at the Thirteenth International Conference on Learning Representations (ICLR 2025)Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
Self-supervised learning (SSL) approaches have brought tremendous success across many tasks and domains. It has been argued that these successes can be attributed to a link between SSL and identifiable representation learning: Temporal structure and auxiliary variables ensure that latent representations are related to the true underlying generative factors of the data. Here, we deepen this connection and show that SSL can perform system identification in latent space. We propose dynamics contrastive learning, a framework to uncover linear, switching linear and non-linear dynamics under a non-linear observation model, give theoretical guarantees and validate them empirically.
- [620] arXiv:2410.20445 (replaced) [pdf, html, other]
-
Title: TrajAgent: An LLM-based Agent Framework for Automated Trajectory Modeling via Collaboration of Large and Small ModelsComments: the code will be openly accessible at: this https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Trajectory modeling, which includes research on trajectory data pattern mining and future prediction, has widespread applications in areas such as life services, urban transportation, and public administration. Numerous methods have been proposed to address specific problems within trajectory modeling. However, the heterogeneity of data and the diversity of trajectory tasks make effective and reliable trajectory modeling an important yet highly challenging endeavor, even for domain experts. In this paper, we propose \textit{TrajAgent}, a agent framework powered by large language models (LLMs), designed to facilitate robust and efficient trajectory modeling through automation modeling. This framework leverages and optimizes diverse specialized models to address various trajectory modeling tasks across different datasets effectively. In \textit{TrajAgent}, we first develop \textit{UniEnv}, an execution environment with a unified data and model interface, to support the execution and training of various models. Building on \textit{UniEnv}, we introduce an agentic workflow designed for automatic trajectory modeling across various trajectory tasks and data. Furthermore, we introduce collaborative learning schema between LLM-based agents and small speciallized models, to enhance the performance of the whole framework effectively. Extensive experiments on four tasks using four real-world datasets demonstrate the effectiveness of \textit{TrajAgent} in automated trajectory modeling, achieving a performance improvement of 2.38\%-34.96\% over baseline methods.
- [621] arXiv:2410.21119 (replaced) [pdf, html, other]
-
Title: A Unified Solution to Diverse Heterogeneities in One-shot Federated LearningComments: Accepted version to KDD 2025Subjects: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
One-Shot Federated Learning (OSFL) restricts communication between the server and clients to a single round, significantly reducing communication costs and minimizing privacy leakage risks compared to traditional Federated Learning (FL), which requires multiple rounds of communication. However, existing OSFL frameworks remain vulnerable to distributional heterogeneity, as they primarily focus on model heterogeneity while neglecting data heterogeneity. To bridge this gap, we propose FedHydra, a unified, data-free, OSFL framework designed to effectively address both model and data heterogeneity. Unlike existing OSFL approaches, FedHydra introduces a novel two-stage learning mechanism. Specifically, it incorporates model stratification and heterogeneity-aware stratified aggregation to mitigate the challenges posed by both model and data heterogeneity. By this design, the data and model heterogeneity issues are simultaneously monitored from different aspects during learning. Consequently, FedHydra can effectively mitigate both issues by minimizing their inherent conflicts. We compared FedHydra with five SOTA baselines on four benchmark datasets. Experimental results show that our method outperforms the previous OSFL methods in both homogeneous and heterogeneous settings. The code is available at this https URL.
- [622] arXiv:2411.00355 (replaced) [pdf, html, other]
-
Title: TextDestroyer: A Training- and Annotation-Free Diffusion Method for Destroying Anomal Text from ImagesSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
In this paper, we propose TextDestroyer, the first training- and annotation-free method for scene text destruction using a pre-trained diffusion model. Existing scene text removal models require complex annotation and retraining, and may leave faint yet recognizable text information, compromising privacy protection and content concealment. TextDestroyer addresses these issues by employing a three-stage hierarchical process to obtain accurate text masks. Our method scrambles text areas in the latent start code using a Gaussian distribution before reconstruction. During the diffusion denoising process, self-attention key and value are referenced from the original latent to restore the compromised background. Latent codes saved at each inversion step are used for replacement during reconstruction, ensuring perfect background restoration. The advantages of TextDestroyer include: (1) it eliminates labor-intensive data annotation and resource-intensive training; (2) it achieves more thorough text destruction, preventing recognizable traces; and (3) it demonstrates better generalization capabilities, performing well on both real-world scenes and generated images.
- [623] arXiv:2411.04794 (replaced) [pdf, html, other]
-
Title: KnowCoder-X: Boosting Multilingual Information Extraction via CodeYuxin Zuo, Wenxuan Jiang, Wenxuan Liu, Zixuan Li, Long Bai, Hanbin Wang, Yutao Zeng, Xiaolong Jin, Jiafeng Guo, Xueqi ChengComments: ACL 2025 FindingsSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Empirical evidence indicates that LLMs exhibit spontaneous cross-lingual alignment. However, although LLMs show promising cross-lingual alignment in Information Extraction (IE), a significant imbalance across languages persists, highlighting an underlying deficiency. To address this, we propose KnowCoder-X, a powerful code LLM with advanced cross-lingual and multilingual capabilities for universal IE. Firstly, it standardizes the representation of multilingual schemas using Python classes, ensuring a consistent ontology across different languages. Then, IE across languages is formulated as a unified code generation task. Secondly, we conduct IE cross-lingual alignment instruction tuning on the translated instance prediction task to enhance the model's cross-lingual transferability. During this phase, we also construct a high-quality and diverse bilingual IE parallel dataset with 257k samples, called ParallelNER, synthesized by our proposed robust three-stage pipeline, with manual annotation to ensure quality. Although without training in 29 unseen languages, KnowCoder-X surpasses ChatGPT by 30.17\% and SoTA by 20.03\%, thereby demonstrating superior cross-lingual IE capabilities. Comprehensive evaluations on 64 IE benchmarks in Chinese and English under various settings demonstrate that KnowCoder-X significantly enhances cross-lingual IE transfer through boosting the IE alignment. Our code and dataset are available at: this https URL
- [624] arXiv:2411.05980 (replaced) [pdf, html, other]
-
Title: FactLens: Benchmarking Fine-Grained Fact VerificationComments: 12 pages, updated versionSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Large Language Models (LLMs) have shown impressive capability in language generation and understanding, but their tendency to hallucinate and produce factually incorrect information remains a key limitation. To verify LLM-generated contents and claims from other sources, traditional verification approaches often rely on holistic models that assign a single factuality label to complex claims, potentially obscuring nuanced errors. In this paper, we advocate for a shift towards fine-grained verification, where complex claims are broken down into smaller sub-claims for individual verification, allowing for more precise identification of inaccuracies, improved transparency, and reduced ambiguity in evidence retrieval. However, generating sub-claims poses challenges, such as maintaining context and ensuring semantic equivalence with respect to the original claim. We introduce FactLens, a benchmark for evaluating fine-grained fact verification, with metrics and automated evaluators of sub-claim quality. The benchmark data is manually curated to ensure high-quality ground truth. Our results show alignment between automated FactLens evaluators and human judgments, and we discuss the impact of sub-claim characteristics on the overall verification performance.
- [625] arXiv:2411.18688 (replaced) [pdf, html, other]
-
Title: Immune: Improving Safety Against Jailbreaks in Multi-modal LLMs via Inference-Time AlignmentSoumya Suvra Ghosal, Souradip Chakraborty, Vaibhav Singh, Tianrui Guan, Mengdi Wang, Alvaro Velasquez, Ahmad Beirami, Furong Huang, Dinesh Manocha, Amrit Singh BediComments: Accepted to CVPR 2025Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
With the widespread deployment of Multimodal Large Language Models (MLLMs) for visual-reasoning tasks, improving their safety has become crucial. Recent research indicates that despite training-time safety alignment, these models remain vulnerable to jailbreak attacks. In this work, we first highlight an important safety gap to describe that alignment achieved solely through safety training may be insufficient against jailbreak attacks. To address this vulnerability, we propose Immune, an inference-time defense framework that leverages a safe reward model through controlled decoding to defend against jailbreak attacks. Additionally, we provide a mathematical characterization of Immune, offering insights on why it improves safety against jailbreaks. Extensive evaluations on diverse jailbreak benchmarks using recent MLLMs reveal that Immune effectively enhances model safety while preserving the model's original capabilities. For instance, against text-based jailbreak attacks on LLaVA-1.6, Immune reduces the attack success rate by 57.82% and 16.78% compared to the base MLLM and state-of-the-art defense strategy, respectively.
- [626] arXiv:2411.18964 (replaced) [pdf, other]
-
Title: Neural Operators for Predictor Feedback Control of Nonlinear Delay SystemsComments: 26 pages. Learning for Dynamics and Control 2025Subjects: Systems and Control (eess.SY); Machine Learning (cs.LG); Dynamical Systems (math.DS); Optimization and Control (math.OC)
Predictor feedback designs are critical for delay-compensating controllers in nonlinear systems. However, these designs are limited in practical applications as predictors cannot be directly implemented, but require numerical approximation schemes, which become computationally prohibitive when system dynamics are expensive to compute. To address this challenge, we recast the predictor design as an operator learning problem, and learn the predictor mapping via a neural operator. We prove the existence of an arbitrarily accurate neural operator approximation of the predictor operator. Under the approximated predictor, we achieve semiglobal practical stability of the closed-loop nonlinear delay system. The estimate is semiglobal in a unique sense - one can enlarge the set of initial states as desired, though this increases the difficulty of training a neural operator, which appears practically in the stability estimate. Furthermore, our analysis holds for any black-box predictor satisfying the universal approximation error bound. We demonstrate the approach by controlling a 5-link robotic manipulator with different neural operator models, achieving significant speedups compared to classic predictor feedback schemes while maintaining closed-loop stability.
- [627] arXiv:2412.03238 (replaced) [pdf, html, other]
-
Title: Dynamic Consistent $k$-Center Clustering with Optimal RecourseComments: In the Proceedings of SODA 2025Subjects: Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG)
Given points from an arbitrary metric space and a sequence of point updates sent by an adversary, what is the minimum recourse per update (i.e., the minimum number of changes needed to the set of centers after an update), in order to maintain a constant-factor approximation to a $k$-clustering problem? This question has received attention in recent years under the name consistent clustering.
Previous works by Lattanzi and Vassilvitskii [ICLM '17] and Fichtenberger, Lattanzi, Norouzi-Fard, and Svensson [SODA '21] studied $k$-clustering objectives, including the $k$-center and the $k$-median objectives, under only point insertions. In this paper we study the $k$-center objective in the fully dynamic setting, where the update is either a point insertion or a point deletion. Before our work, Łącki, Haeupler, Grunau, Rozhoň, and Jayaram [SODA '24] gave a deterministic fully dynamic constant-factor approximation algorithm for the $k$-center objective with worst-case recourse of $2$ per update.
In this work, we prove that the $k$-center clustering problem admits optimal recourse bounds by developing a deterministic fully dynamic constant-factor approximation algorithm with worst-case recourse of $1$ per update. Moreover our algorithm performs simple choices based on light data structures, and thus is arguably more direct and faster than the previous one which uses a sophisticated combinatorial structure. Additionally, we develop a new deterministic decremental algorithm and a new deterministic incremental algorithm, both of which maintain a $6$-approximate $k$-center solution with worst-case recourse of $1$ per update. Our incremental algorithm improves over the $8$-approximation algorithm by Charikar, Chekuri, Feder, and Motwani [STOC '97]. Finally, we remark that since all three of our algorithms are deterministic, they work against an adaptive adversary. - [628] arXiv:2412.03795 (replaced) [pdf, html, other]
-
Title: Samudra: An AI Global Ocean Emulator for ClimateSurya Dheeshjith, Adam Subel, Alistair Adcroft, Julius Busecke, Carlos Fernandez-Granda, Shubham Gupta, Laure ZannaJournal-ref: Geophysical Research Letters 52.10 (2025)Subjects: Atmospheric and Oceanic Physics (physics.ao-ph); Machine Learning (cs.LG)
AI emulators for forecasting have emerged as powerful tools that can outperform conventional numerical predictions. The next frontier is to build emulators for long climate simulations with skill across a range of spatiotemporal scales, a particularly important goal for the ocean. Our work builds a skillful global emulator of the ocean component of a state-of-the-art climate model. We emulate key ocean variables, sea surface height, horizontal velocities, temperature, and salinity, across their full depth. We use a modified ConvNeXt UNet architecture trained on multi-depth levels of ocean data. We show that the ocean emulator - Samudra - which exhibits no drift relative to the truth, can reproduce the depth structure of ocean variables and their interannual variability. Samudra is stable for centuries and 150 times faster than the original ocean model. Samudra struggles to capture the correct magnitude of the forcing trends and simultaneously remain stable, requiring further work.
- [629] arXiv:2412.05488 (replaced) [pdf, html, other]
-
Title: Enhancing Sample Generation of Diffusion Models using Noise Level CorrectionJournal-ref: Transactions on Machine Learning Research, 2025Subjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
The denoising process of diffusion models can be interpreted as an approximate projection of noisy samples onto the data manifold. Moreover, the noise level in these samples approximates their distance to the underlying manifold. Building on this insight, we propose a novel method to enhance sample generation by aligning the estimated noise level with the true distance of noisy samples to the manifold. Specifically, we introduce a noise level correction network, leveraging a pre-trained denoising network, to refine noise level estimates during the denoising process. Additionally, we extend this approach to various image restoration tasks by integrating task-specific constraints, including inpainting, deblurring, super-resolution, colorization, and compressed sensing. Experimental results demonstrate that our method significantly improves sample quality in both unconstrained and constrained generation scenarios. Notably, the proposed noise level correction framework is compatible with existing denoising schedulers (e.g., DDIM), offering additional performance improvements.
- [630] arXiv:2412.05718 (replaced) [pdf, html, other]
-
Title: RLZero: Direct Policy Inference from Language Without In-Domain SupervisionHarshit Sikchi, Siddhant Agarwal, Pranaya Jajoo, Samyak Parajuli, Caleb Chuck, Max Rudolph, Peter Stone, Amy Zhang, Scott NiekumComments: 26 pagesSubjects: Artificial Intelligence (cs.AI); Graphics (cs.GR); Machine Learning (cs.LG); Robotics (cs.RO)
The reward hypothesis states that all goals and purposes can be understood as the maximization of a received scalar reward signal. However, in practice, defining such a reward signal is notoriously difficult, as humans are often unable to predict the optimal behavior corresponding to a reward function. Natural language offers an intuitive alternative for instructing reinforcement learning (RL) agents, yet previous language-conditioned approaches either require costly supervision or test-time training given a language instruction. In this work, we present a new approach that uses a pretrained RL agent trained using only unlabeled, offline interactions--without task-specific supervision or labeled trajectories--to get zero-shot test-time policy inference from arbitrary natural language instructions. We introduce a framework comprising three steps: imagine, project, and imitate. First, the agent imagines a sequence of observations corresponding to the provided language description using video generative models. Next, these imagined observations are projected into the target environment domain. Finally, an agent pretrained in the target environment with unsupervised RL instantly imitates the projected observation sequence through a closed-form solution. To the best of our knowledge, our method, RLZero, is the first approach to show direct language-to-behavior generation abilities on a variety of tasks and environments without any in-domain supervision. We further show that components of RLZero can be used to generate policies zero-shot from cross-embodied videos, such as those available on YouTube, even for complex embodiments like humanoids.
- [631] arXiv:2412.11569 (replaced) [pdf, html, other]
-
Title: The dark side of the forces: assessing non-conservative force models for atomistic machine learningComments: 10 pages (including references) + appendix Conference formatSubjects: Chemical Physics (physics.chem-ph); Machine Learning (cs.LG)
The use of machine learning to estimate the energy of a group of atoms, and the forces that drive them to more stable configurations, has revolutionized the fields of computational chemistry and materials discovery. In this domain, rigorous enforcement of symmetry and conservation laws has traditionally been considered essential. For this reason, interatomic forces are usually computed as the derivatives of the potential energy, ensuring energy conservation. Several recent works have questioned this physically constrained approach, suggesting that directly predicting the forces yields a better trade-off between accuracy and computational efficiency -- and that energy conservation can be learned during training. This work investigates the applicability of such non-conservative models in microscopic simulations. We identify and demonstrate several fundamental issues, from ill-defined convergence of geometry optimization to instability in various types of molecular dynamics. Contrary to the case of rotational symmetry, energy conservation is hard to learn, monitor, and correct for. The best approach to exploit the acceleration afforded by direct force prediction might be to use it in tandem with a conservative model, reducing -- rather than eliminating -- the additional cost of backpropagation, but avoiding the pathological behavior associated with non-conservative forces.
- [632] arXiv:2412.11743 (replaced) [pdf, html, other]
-
Title: Generalized Bayesian deep reinforcement learningSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
Bayesian reinforcement learning (BRL) is a method that merges principles from Bayesian statistics and reinforcement learning to make optimal decisions in uncertain environments. As a model-based RL method, it has two key components: (1) inferring the posterior distribution of the model for the data-generating process (DGP) and (2) policy learning using the learned posterior. We propose to model the dynamics of the unknown environment through deep generative models, assuming Markov dependence. In the absence of likelihood functions for these models, we train them by learning a generalized predictive-sequential (or prequential) scoring rule (SR) posterior. We used sequential Monte Carlo (SMC) samplers to draw samples from this generalized Bayesian posterior distribution. In conjunction, to achieve scalability in the high-dimensional parameter space of the neural networks, we use the gradient-based Markov kernels within SMC. To justify the use of the prequential scoring rule posterior, we prove a Bernstein-von Mises-type theorem. For policy learning, we propose expected Thompson sampling (ETS) to learn the optimal policy by maximising the expected value function with respect to the posterior distribution. This improves upon traditional Thompson sampling (TS) and its extensions, which utilize only one sample drawn from the posterior distribution. This improvement is studied both theoretically and using simulation studies, assuming a discrete action space. Finally, we successfully extended our setup for a challenging problem with a continuous action space without theoretical guarantees.
- [633] arXiv:2412.12276 (replaced) [pdf, html, other]
-
Title: Emergence and Effectiveness of Task Vectors in In-Context Learning: An Encoder Decoder PerspectiveComments: this https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Autoregressive transformers exhibit adaptive learning through in-context learning (ICL), which begs the question of how. Prior works have shown that transformers represent the ICL tasks as vectors in their representations. In this paper, we leverage the encoding-decoding framework to study how transformers form task vectors during pretraining and how their task encoding quality predicts ICL task performance. On synthetic ICL tasks, we analyze the training dynamics of a small transformer and report the coupled emergence of task encoding and decoding. As the model learns to encode different latent tasks (e.g., "Finding the first noun in a sentence.") into distinct, separable representations, it concurrently builds conditional decoding algorithms and improves its ICL performance. We validate this phenomenon across pretrained models of varying scales (Gemma-2 2B/9B/27B, Llama-3.1 8B/70B) and over the course of pretraining in OLMo-7B. Further, we demonstrate that the quality of task encoding inferred from representations predicts ICL performance, and that, surprisingly, finetuning the earlier layers can improve the task encoding and performance more than finetuning the latter layers. Our empirical insights shed light into better understanding the success and failure modes of large language models via their representations.
- [634] arXiv:2412.12987 (replaced) [pdf, html, other]
-
Title: Stochastic interior-point methods for smooth conic optimization with applicationsSubjects: Optimization and Control (math.OC); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Conic optimization plays a crucial role in many machine learning (ML) problems. However, practical algorithms for conic constrained ML problems with large datasets are often limited to specific use cases, as stochastic algorithms for general conic optimization remain underdeveloped. To fill this gap, we introduce a stochastic interior-point method (SIPM) framework for general conic optimization, along with four novel SIPM variants leveraging distinct stochastic gradient estimators. Under mild assumptions, we establish the iteration complexity of our proposed SIPMs, which, up to a polylogarithmic factor, match the best-known results in stochastic unconstrained optimization. Finally, our numerical experiments on robust linear regression, multi-task relationship learning, and clustering data streams demonstrate the effectiveness and efficiency of our approach.
- [635] arXiv:2412.18277 (replaced) [pdf, html, other]
-
Title: Towards Modality Generalization: A Benchmark and Prospective AnalysisComments: under-reviewSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Multi-modal learning has achieved remarkable success by integrating information from various modalities, achieving superior performance in tasks like recognition and retrieval compared to uni-modal approaches. However, real-world scenarios often present novel modalities that are unseen during training due to resource and privacy constraints, a challenge current methods struggle to address. This paper introduces Modality Generalization (MG), which focuses on enabling models to generalize to unseen modalities. We define two cases: Weak MG, where both seen and unseen modalities can be mapped into a joint embedding space via existing perceptors, and Strong MG, where no such mappings exist. To facilitate progress, we propose a comprehensive benchmark featuring multi-modal algorithms and adapt existing methods that focus on generalization. Extensive experiments highlight the complexity of MG, exposing the limitations of existing methods and identifying key directions for future research. Our work provides a foundation for advancing robust and adaptable multi-modal models, enabling them to handle unseen modalities in realistic scenarios.
- [636] arXiv:2412.18547 (replaced) [pdf, html, other]
-
Title: Token-Budget-Aware LLM ReasoningSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Reasoning is critical for large language models (LLMs) to excel in a wide range of tasks. While methods like Chain-of-Thought (CoT) reasoning and enhance LLM performance by decomposing problems into intermediate steps, they also incur significant overhead in token usage, leading to increased costs. We find that the reasoning process of current LLMs is unnecessarily lengthy and it can be compressed by including a reasonable token budget in the prompt, but the choice of token budget plays a crucial role in the actual compression effectiveness. We then propose a token-budget-aware LLM reasoning framework that dynamically adjusts the number of reasoning tokens based on the reasoning complexity of each problem. Experiments show that our method effectively reduces token costs in CoT reasoning with only a slight performance reduction, offering a practical solution to balance efficiency and accuracy in LLM reasoning. Code: this https URL
- [637] arXiv:2412.20070 (replaced) [pdf, html, other]
-
Title: Exploring Compositional Generalization of Multimodal LLMs for Medical ImagingZhenyang Cai, Junying Chen, Rongsheng Wang, Weihong Wang, Yonglin Deng, Dingjie Song, Yize Chen, Zixu Zhang, Benyou WangSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Medical imaging provides essential visual insights for diagnosis, and multimodal large language models (MLLMs) are increasingly utilized for its analysis due to their strong generalization capabilities; however, the underlying factors driving this generalization remain unclear. Current research suggests that multi-task training outperforms single-task as different tasks can benefit each other, but they often overlook the internal relationships within these tasks. To analyze this phenomenon, we attempted to employ compositional generalization (CG), which refers to the models' ability to understand novel combinations by recombining learned elements, as a guiding framework. Since medical images can be precisely defined by Modality, Anatomical area, and Task, naturally providing an environment for exploring CG, we assembled 106 medical datasets to create Med-MAT for comprehensive experiments. The experiments confirmed that MLLMs can use CG to understand unseen medical images and identified CG as one of the main drivers of the generalization observed in multi-task training. Additionally, further studies demonstrated that CG effectively supports datasets with limited data and confirmed that MLLMs can achieve CG across classification and detection tasks, underscoring its broader generalization potential. Med-MAT is available at this https URL.
- [638] arXiv:2501.01377 (replaced) [pdf, html, other]
-
Title: Improving Medical Large Vision-Language Models with Abnormal-Aware FeedbackComments: 16 pagesSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Existing Medical Large Vision-Language Models (Med-LVLMs), encapsulating extensive medical knowledge, demonstrate excellent capabilities in understanding medical images. However, there remain challenges in visual localization in medical images, which is crucial for abnormality detection and interpretation. To address these issues, we propose a novel UMed-LVLM designed to unveil medical abnormalities. Specifically, we collect a Medical Abnormalities Unveiling (MAU) dataset and propose a two-stage training method for UMed-LVLM training. To collect MAU dataset, we propose a prompt method utilizing the GPT-4V to generate diagnoses based on identified abnormal areas in medical images. Moreover, the two-stage training method includes Abnormal-Aware Instruction Tuning and Abnormal-Aware Rewarding, comprising Relevance Reward, Abnormal Localization Reward and Vision Relevance Reward. Experimental results demonstrate that our UMed-LVLM significantly outperforms existing Med-LVLMs in identifying and understanding medical abnormalities, achieving a 58% improvement over the baseline. In addition, this work shows that enhancing the abnormality detection capabilities of Med-LVLMs significantly improves their understanding of medical images and generalization capability.
- [639] arXiv:2501.01710 (replaced) [pdf, html, other]
-
Title: Enhancing Large Vision Model in Street Scene Semantic Understanding through Leveraging Posterior Optimization TrajectoryWei-Bin Kou, Qingfeng Lin, Ming Tang, Jingreng Lei, Shuai Wang, Rongguang Ye, Guangxu Zhu, Yik-Chung WuComments: 7 pagesSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Robotics (cs.RO)
To improve the generalization of the autonomous driving (AD) perception model, vehicles need to update the model over time based on the continuously collected data. As time progresses, the amount of data fitted by the AD model expands, which helps to improve the AD model generalization substantially. However, such ever-expanding data is a double-edged sword for the AD model. Specifically, as the fitted data volume grows to exceed the the AD model's fitting capacities, the AD model is prone to under-fitting. To address this issue, we propose to use a pretrained Large Vision Models (LVMs) as backbone coupled with downstream perception head to understand AD semantic information. This design can not only surmount the aforementioned under-fitting problem due to LVMs' powerful fitting capabilities, but also enhance the perception generalization thanks to LVMs' vast and diverse training data. On the other hand, to mitigate vehicles' computational burden of training the perception head while running LVM backbone, we introduce a Posterior Optimization Trajectory (POT)-Guided optimization scheme (POTGui) to accelerate the convergence. Concretely, we propose a POT Generator (POTGen) to generate posterior (future) optimization direction in advance to guide the current optimization iteration, through which the model can generally converge within 10 epochs. Extensive experiments demonstrate that the proposed method improves the performance by over 66.48\% and converges faster over 6 times, compared to the existing state-of-the-art approach.
- [640] arXiv:2501.01999 (replaced) [pdf, html, other]
-
Title: Probing Equivariance and Symmetry Breaking in Convolutional NetworksSharvaree Vadgama, Mohammad Mohaiminul Islam, Domas Buracas, Christian Shewmake, Artem Moskalev, Erik BekkersComments: 27 pages, 7 figuresSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
In this work, we explore the trade-offs of explicit structural priors, particularly group equivariance. We address this through theoretical analysis and a comprehensive empirical study. To enable controlled and fair comparisons, we introduce \texttt{Rapidash}, a unified group convolutional architecture that allows for different variants of equivariant and non-equivariant models. Our results suggest that more constrained equivariant models outperform less constrained alternatives when aligned with the geometry of the task, and increasing representation capacity does not fully eliminate performance gaps. We see improved performance of models with equivariance and symmetry-breaking through tasks like segmentation, regression, and generation across diverse datasets. Explicit \textit{symmetry breaking} via geometric reference frames consistently improves performance, while \textit{breaking equivariance} through geometric input features can be helpful when aligned with task geometry. Our results provide task-specific performance trends that offer a more nuanced way for model selection.
- [641] arXiv:2501.02669 (replaced) [pdf, html, other]
-
Title: Generalizing from SIMPLE to HARD Visual Reasoning: Can We Mitigate Modality Imbalance in VLMs?Subjects: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL); Machine Learning (cs.LG)
Vision Language Models (VLMs) are impressive at visual question answering and image captioning. But they underperform on multi-step visual reasoning -- even compared to LLMs on the same tasks presented in text form -- giving rise to perceptions of modality imbalance or brittleness. Towards a systematic study of such issues, we introduce a synthetic framework for assessing the ability of VLMs to perform algorithmic visual reasoning, comprising three tasks: Table Readout, Grid Navigation, and Visual Analogy. Each has two levels of difficulty, SIMPLE and HARD, and even the SIMPLE versions are difficult for frontier VLMs. We propose strategies for training on the SIMPLE version of tasks that improve performance on the corresponding HARD task, i.e., simple-to-hard (S2H) generalization. This controlled setup, where each task also has an equivalent text-only version, allows a quantification of the modality imbalance and how it is impacted by training strategy. We show that 1) explicit image-to-text conversion is important in promoting S2H generalization on images, by transferring reasoning from text; 2) conversion can be internalized at test time. We also report results of mechanistic study of this phenomenon. We identify measures of gradient alignment that can identify training strategies that promote better S2H generalization. Ablations highlight the importance of chain-of-thought.
- [642] arXiv:2501.04292 (replaced) [pdf, html, other]
-
Title: MADUV: The 1st INTERSPEECH Mice Autism Detection via Ultrasound Vocalization ChallengeZijiang Yang, Meishu Song, Xin Jing, Haojie Zhang, Kun Qian, Bin Hu, Kota Tamada, Toru Takumi, Björn W. Schuller, Yoshiharu YamamotoComments: 5 pages, 1 figure and 2 tables. Submitted to INTERSPEECH 2025. For MADUV Challenge 2025Subjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
The Mice Autism Detection via Ultrasound Vocalization (MADUV) Challenge introduces the first INTERSPEECH challenge focused on detecting autism spectrum disorder (ASD) in mice through their vocalizations. Participants are tasked with developing models to automatically classify mice as either wild-type or ASD models based on recordings with a high sampling rate. Our baseline system employs a simple CNN-based classification using three different spectrogram features. Results demonstrate the feasibility of automated ASD detection, with the considered audible-range features achieving the best performance (UAR of 0.600 for segment-level and 0.625 for subject-level classification). This challenge bridges speech technology and biomedical research, offering opportunities to advance our understanding of ASD models through machine learning approaches. The findings suggest promising directions for vocalization analysis and highlight the potential value of audible and ultrasound vocalizations in ASD detection.
- [643] arXiv:2501.05966 (replaced) [pdf, html, other]
-
Title: Towards Early Prediction of Self-Supervised Speech Model PerformanceSubjects: Sound (cs.SD); Computation and Language (cs.CL); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
In Self-Supervised Learning (SSL), pre-training and evaluation are resource intensive. In the speech domain, current indicators of the quality of SSL models during pre-training, such as the loss, do not correlate well with downstream performance. Consequently, it is often difficult to gauge the final downstream performance in a cost efficient manner during pre-training. In this work, we propose unsupervised efficient methods that give insights into the quality of the pre-training of SSL speech models, namely, measuring the cluster quality and rank of the embeddings of the SSL model. Results show that measures of cluster quality and rank correlate better with downstream performance than the pre-training loss with only one hour of unlabeled audio, reducing the need for GPU hours and labeled data in SSL model evaluation.
- [644] arXiv:2501.07985 (replaced) [pdf, html, other]
-
Title: CHEQ-ing the Box: Safe Variable Impedance Learning for Robotic PolishingSubjects: Robotics (cs.RO); Machine Learning (cs.LG)
Robotic systems are increasingly employed for industrial automation, with contact-rich tasks like polishing requiring dexterity and compliant behaviour. These tasks are difficult to model, making classical control challenging. Deep reinforcement learning (RL) offers a promising solution by enabling the learning of models and control policies directly from data. However, its application to real-world problems is limited by data inefficiency and unsafe exploration. Adaptive hybrid RL methods blend classical control and RL adaptively, combining the strengths of both: structure from control and learning from RL. This has led to improvements in data efficiency and exploration safety. However, their potential for hardware applications remains underexplored, with no evaluations on physical systems to date. Such evaluations are critical to fully assess the practicality and effectiveness of these methods in real-world settings. This work presents an experimental demonstration of the hybrid RL algorithm CHEQ for robotic polishing with variable impedance, a task requiring precise force and velocity tracking. In simulation, we show that variable impedance enhances polishing performance. We compare standalone RL with adaptive hybrid RL, demonstrating that CHEQ achieves effective learning while adhering to safety constraints. On hardware, CHEQ achieves effective polishing behaviour, requiring only eight hours of training and incurring just five failures. These results highlight the potential of adaptive hybrid RL for real-world, contact-rich tasks trained directly on hardware.
- [645] arXiv:2501.13125 (replaced) [pdf, html, other]
-
Title: Generating Plausible Distractors for Multiple-Choice Questions via Student Choice PredictionComments: This paper has been accepted for publication at ACL 2025Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
In designing multiple-choice questions (MCQs) in education, creating plausible distractors is crucial for identifying students' misconceptions and gaps in knowledge and accurately assessing their understanding. However, prior studies on distractor generation have not paid sufficient attention to enhancing the difficulty of distractors, resulting in reduced effectiveness of MCQs. This study presents a pipeline for training a model to generate distractors that are more likely to be selected by students. First, we train a pairwise ranker to reason about students' misconceptions and assess the relative plausibility of two distractors. Using this model, we create a dataset of pairwise distractor ranks and then train a distractor generator via Direct Preference Optimization (DPO) to generate more plausible distractors. Experiments on computer science subjects (Python, DB, MLDL) demonstrate that our pairwise ranker effectively identifies students' potential misunderstandings and achieves ranking accuracy comparable to human experts. Furthermore, our distractor generator outperforms several baselines in generating plausible distractors and produces questions with a higher item discrimination index (DI).
- [646] arXiv:2501.13134 (replaced) [pdf, html, other]
-
Title: UniRestore: Unified Perceptual and Task-Oriented Image Restoration Model Using Diffusion PriorComments: Accepted by CVPR2025 (Highlight); Project Page: this https URLSubjects: Image and Video Processing (eess.IV); Machine Learning (cs.LG)
Image restoration aims to recover content from inputs degraded by various factors, such as adverse weather, blur, and noise. Perceptual Image Restoration (PIR) methods improve visual quality but often do not support downstream tasks effectively. On the other hand, Task-oriented Image Restoration (TIR) methods focus on enhancing image utility for high-level vision tasks, sometimes compromising visual quality. This paper introduces UniRestore, a unified image restoration model that bridges the gap between PIR and TIR by using a diffusion prior. The diffusion prior is designed to generate images that align with human visual quality preferences, but these images are often unsuitable for TIR scenarios. To solve this limitation, UniRestore utilizes encoder features from an autoencoder to adapt the diffusion prior to specific tasks. We propose a Complementary Feature Restoration Module (CFRM) to reconstruct degraded encoder features and a Task Feature Adapter (TFA) module to facilitate adaptive feature fusion in the decoder. This design allows UniRestore to optimize images for both human perception and downstream task requirements, addressing discrepancies between visual quality and functional needs. Integrating these modules also enhances UniRestore's adapability and efficiency across diverse tasks. Extensive expertments demonstrate the superior performance of UniRestore in both PIR and TIR scenarios.
- [647] arXiv:2501.13772 (replaced) [pdf, html, other]
-
Title: Jailbreak-AudioBench: In-Depth Evaluation and Analysis of Jailbreak Threats for Large Audio Language ModelsHao Cheng, Erjia Xiao, Jing Shao, Yichi Wang, Le Yang, Chao Shen, Philip Torr, Jindong Gu, Renjing XuSubjects: Sound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multimedia (cs.MM); Audio and Speech Processing (eess.AS)
Large Language Models (LLMs) demonstrate impressive zero-shot performance across a wide range of natural language processing tasks. Integrating various modality encoders further expands their capabilities, giving rise to Multimodal Large Language Models (MLLMs) that process not only text but also visual and auditory modality inputs. However, these advanced capabilities may also pose significant security risks, as models can be exploited to generate harmful or inappropriate content through jailbreak attack. While prior work has extensively explored how manipulating textual or visual modality inputs can circumvent safeguards in LLMs and MLLMs, the vulnerability of audio-specific Jailbreak on Large Audio-Language Models (LALMs) remains largely underexplored. To address this gap, we introduce \textbf{Jailbreak-AudioBench}, which consists of the Toolbox, curated Dataset, and comprehensive Benchmark. The Toolbox supports not only text-to-audio conversion but also various editing techniques for injecting audio hidden semantics. The curated Dataset provides diverse explicit and implicit jailbreak audio examples in both original and edited forms. Utilizing this dataset, we evaluate multiple state-of-the-art LALMs and establish the most comprehensive Jailbreak benchmark to date for audio modality. Finally, Jailbreak-AudioBench establishes a foundation for advancing future research on LALMs safety alignment by enabling the in-depth exposure of more powerful jailbreak threats, such as query-based audio editing, and by facilitating the development of effective defense mechanisms.
- [648] arXiv:2501.15056 (replaced) [pdf, html, other]
-
Title: Feedback-Aware Monte Carlo Tree Search for Efficient Information Seeking in Goal-Oriented ConversationsSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
Effective decision-making and problem-solving in conversational systems require the ability to identify and acquire missing information through targeted questioning. A key challenge lies in efficiently narrowing down a large space of possible outcomes by posing questions that minimize uncertainty. To address this, we introduce a novel framework that leverages Large Language Models (LLMs) to generate information-seeking questions, with Monte Carlo Tree Search (MCTS) to strategically select questions that maximize information gain, as a part of inference-time planning. Our primary contribution includes a hierarchical feedback mechanism that exploits past interaction patterns to guide future strategy. Specifically, each new problem is mapped to a cluster based on semantic similarity, and our UCT (Upper Confidence bound for Trees) formulation employs a cluster-specific bonus reward to prioritize successful question trajectories that have proven effective for similar problems in the past. Extensive empirical evaluation across medical diagnosis and technical troubleshooting domains shows that our method achieves an average of 12% improvement in success rates and about 10x reduction in the number of LLM calls made for planning per conversation, compared to the state of the art. An additional 8% gain in success rate is observed on average when we start with a constrained set of possibilities. Our results underscore the efficacy of feedback-aware MCTS in enhancing information-seeking in goal-oriented dialogues.
- [649] arXiv:2501.17823 (replaced) [pdf, html, other]
-
Title: Robust Multimodal Learning via Cross-Modal Proxy TokensComments: 21 Pages, 9 Figures, 6 TablesSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Multimodal models often experience a significant performance drop when one or more modalities are missing during inference. To address this challenge, we propose a simple yet effective approach that enhances robustness to missing modalities while maintaining strong performance when all modalities are available. Our method introduces cross-modal proxy tokens (CMPTs), which approximate the class token of a missing modality by attending only to the tokens of the available modality without requiring explicit modality generation or auxiliary networks. To efficiently learn these approximations with minimal computational overhead, we employ low-rank adapters in frozen unimodal encoders and jointly optimize an alignment loss with a task-specific loss. Extensive experiments on five multimodal datasets show that our method outperforms state-of-the-art baselines across various missing rates while achieving competitive results in complete-modality settings. Overall, our method offers a flexible and efficient solution for robust multimodal learning. The code and pretrained models will be released on GitHub.
- [650] arXiv:2501.18756 (replaced) [pdf, html, other]
-
Title: A Unified Framework for Entropy Search and Expected Improvement in Bayesian OptimizationSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC)
Bayesian optimization is a widely used method for optimizing expensive black-box functions, with Expected Improvement being one of the most commonly used acquisition functions. In contrast, information-theoretic acquisition functions aim to reduce uncertainty about the function's optimum and are often considered fundamentally distinct from EI. In this work, we challenge this prevailing perspective by introducing a unified theoretical framework, Variational Entropy Search, which reveals that EI and information-theoretic acquisition functions are more closely related than previously recognized. We demonstrate that EI can be interpreted as a variational inference approximation of the popular information-theoretic acquisition function, named Max-value Entropy Search. Building on this insight, we propose VES-Gamma, a novel acquisition function that balances the strengths of EI and MES. Extensive empirical evaluations across both low- and high-dimensional synthetic and real-world benchmarks demonstrate that VES-Gamma is competitive with state-of-the-art acquisition functions and in many cases outperforms EI and MES.
- [651] arXiv:2501.18897 (replaced) [pdf, html, other]
-
Title: Statistical Inference for Generative Model ComparisonSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
Generative models have recently achieved remarkable empirical performance in various applications, however, their evaluations yet lack uncertainty quantification. In this paper, we propose a method to compare two generative models with statistical confidence based on an unbiased estimator of their relative performance gap. Theoretically, our estimator achieves parametric convergence rates and admits asymptotic normality, which enables valid inference. Empirically, on simulated datasets, our approach effectively controls type I error without compromising its power. In addition, on real image and language datasets, we demonstrate our method's performance in comparing generative models with statistical guarantees.
- [652] arXiv:2502.03678 (replaced) [pdf, html, other]
-
Title: Reflection-Window Decoding: Text Generation with Selective RefinementZeyu Tang, Zhenhao Chen, Xiangchen Song, Loka Li, Yunlong Deng, Yifan Shen, Guangyi Chen, Peter Spirtes, Kun ZhangComments: In Proceedings of the 42nd International Conference on Machine Learning, 2025. (ICML 2025)Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
The autoregressive decoding for text generation in large language models (LLMs), while widely used, is inherently suboptimal due to the lack of a built-in mechanism to perform refinement and/or correction of the generated content. In this paper, we consider optimality in terms of the joint probability over the generated response, when jointly considering all tokens at the same time. We theoretically characterize the potential deviation of the autoregressively generated response from its globally optimal counterpart that is of the same length. Our analysis suggests that we need to be cautious when noticeable uncertainty arises during text generation, which may signal the sub-optimality of the generation history. To address the pitfall of autoregressive decoding for text generation, we propose an approach that incorporates a sliding reflection window and a pausing criterion, such that refinement and generation can be carried out interchangeably as the decoding proceeds. Our selective refinement framework strikes a balance between efficiency and optimality, and our extensive experimental results demonstrate the effectiveness of our approach.
- [653] arXiv:2502.05449 (replaced) [pdf, html, other]
-
Title: Iterative Deepening Sampling as Efficient Test-Time ScalingSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Recent reasoning models, such as OpenAI's O1 series, have demonstrated exceptional performance on complex reasoning tasks and revealed new test-time scaling laws. Inspired by this, many people have been studying how to train models to achieve effective self-evaluation and self-correction to further enable the scaling paradigm. However, less studied is how to efficiently scale test-time compute from a fixed model, and this remains a challenge. In this paper, we address this challenge by focusing on enhancing the quality of self-reflection data generation for complex problem-solving at test time, which can also subsequently improve the training of next-generation large language models (LLMs). Specifically, we explore how systematically triggering a model's self-correction mechanisms can improve performance on challenging reasoning tasks. To this end, we propose a novel iterative deepening sampling algorithm framework designed to enhance self-correction and generate higher-quality samples. Through extensive experiments on Math500 and AIME benchmarks, we demonstrate that our method achieves a higher success rate on difficult tasks and provide detailed ablation studies to analyze its effectiveness across diverse settings.
- [654] arXiv:2502.06231 (replaced) [pdf, html, other]
-
Title: Falsification of Unconfoundedness by Testing Independence of Causal MechanismsComments: ICML 2025 camera-ready version; 20 pages, including 5 figures, 2 tables, and appendicesSubjects: Methodology (stat.ME); Machine Learning (cs.LG); Machine Learning (stat.ML)
A major challenge in estimating treatment effects in observational studies is the reliance on untestable conditions such as the assumption of no unmeasured confounding. In this work, we propose an algorithm that can falsify the assumption of no unmeasured confounding in a setting with observational data from multiple heterogeneous sources, which we refer to as environments. Our proposed falsification strategy leverages a key observation that unmeasured confounding can cause observed causal mechanisms to appear dependent. Building on this observation, we develop a novel two-stage procedure that detects these dependencies with high statistical power while controlling false positives. The algorithm does not require access to randomized data and, in contrast to other falsification approaches, functions even under transportability violations when the environment has a direct effect on the outcome of interest. To showcase the practical relevance of our approach, we show that our method is able to efficiently detect confounding on both simulated and semi-synthetic data.
- [655] arXiv:2502.11671 (replaced) [pdf, html, other]
-
Title: Diversity-oriented Data Augmentation with Large Language ModelsComments: Accepted to ACL 2025Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Data augmentation is an essential technique in natural language processing (NLP) for enriching training datasets by generating diverse samples. This process is crucial for improving the robustness and generalization capabilities of NLP models. However, a significant challenge remains: \textit{Insufficient Attention to Sample Distribution Diversity}. Most existing methods focus on increasing the sample numbers while neglecting the sample distribution diversity, which can lead to model overfitting. In response, we explore data augmentation's impact on dataset diversity and propose a \textbf{\underline{D}}iversity-\textbf{\underline{o}}riented data \textbf{\underline{Aug}}mentation framework (\textbf{DoAug}). % \(\mathscr{DoAug}\) Specifically, we utilize a diversity-oriented fine-tuning approach to train an LLM as a diverse paraphraser, which is capable of augmenting textual datasets by generating diversified paraphrases. Then, we apply the LLM paraphraser to a selected coreset of highly informative samples and integrate the paraphrases with the original data to create a more diverse augmented dataset. Finally, we conduct extensive experiments on 12 real-world textual datasets. The results show that our fine-tuned LLM augmenter improves diversity while preserving label consistency, thereby enhancing the robustness and performance of downstream tasks. Specifically, it achieves an average performance gain of \(10.52\%\), surpassing the runner-up baseline with more than three percentage points.
- [656] arXiv:2502.13775 (replaced) [pdf, html, other]
-
Title: VITAL: A New Dataset for Benchmarking Pluralistic Alignment in HealthcareComments: Accepted to ACL 2025 (Main Proceedings)Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Alignment techniques have become central to ensuring that Large Language Models (LLMs) generate outputs consistent with human values. However, existing alignment paradigms often model an averaged or monolithic preference, failing to account for the diversity of perspectives across cultures, demographics, and communities. This limitation is particularly critical in health-related scenarios, where plurality is essential due to the influence of culture, religion, personal values, and conflicting opinions. Despite progress in pluralistic alignment, no prior work has focused on health, likely due to the unavailability of publicly available datasets. To address this gap, we introduce VITAL, a new benchmark dataset comprising 13.1K value-laden situations and 5.4K multiple-choice questions focused on health, designed to assess and benchmark pluralistic alignment methodologies. Through extensive evaluation of eight LLMs of varying sizes, we demonstrate that existing pluralistic alignment techniques fall short in effectively accommodating diverse healthcare beliefs, underscoring the need for tailored AI alignment in specific domains. This work highlights the limitations of current approaches and lays the groundwork for developing health-specific alignment solutions.
- [657] arXiv:2502.13943 (replaced) [pdf, html, other]
-
Title: AdaptiveStep: Automatically Dividing Reasoning Step through Model ConfidenceYuliang Liu, Junjie Lu, Zhaoling Chen, Chaofeng Qu, Jason Klein Liu, Chonghan Liu, Zefan Cai, Yunhui Xia, Li Zhao, Jiang Bian, Chuheng Zhang, Wei Shen, Zhouhan LinComments: ICML 2025Subjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Current approaches for training Process Reward Models (PRMs) often involve breaking down responses into multiple reasoning steps using rule-based techniques, such as using predefined placeholder tokens or setting the reasoning step's length into a fixed size. These approaches overlook the fact that specific words do not typically mark true decision points in a text. To address this, we propose AdaptiveStep, a method that divides reasoning steps based on the model's confidence in predicting the next word. This division method provides more decision-making information at each step, enhancing downstream tasks, such as reward model learning. Moreover, our method does not require manual annotation. We demonstrate its effectiveness through experiments with AdaptiveStep-trained PRMs in mathematical reasoning and code generation tasks. Experimental results indicate that the outcome PRM achieves state-of-the-art Best-of-N performance, surpassing greedy search strategy with token-level value-guided decoding, while also reducing construction costs by over 30% compared to existing open-source PRMs. In addition, we provide a thorough analysis and case study on the PRM's performance, transferability, and generalization capabilities.
- [658] arXiv:2502.14060 (replaced) [pdf, html, other]
-
Title: New Lower Bounds for Stochastic Non-Convex Optimization through Divergence DecompositionSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC)
We study fundamental limits of first-order stochastic optimization in a range of nonconvex settings, including L-smooth functions satisfying Quasar-Convexity (QC), Quadratic Growth (QG), and Restricted Secant Inequalities (RSI). While the convergence properties of standard algorithms are well-understood in deterministic regimes, significantly fewer results address the stochastic case, where only unbiased and noisy gradients are available. We establish new lower bounds on the number of noisy gradient queries to minimize these classes of functions, also showing that they are tight (up to a logarithmic factor) in all the relevant quantities characterizing each class. Our approach reformulates the optimization task as a function identification problem, leveraging divergence decomposition arguments to construct a challenging subclass that leads to sharp lower bounds. Furthermore, we present a specialized algorithm in the one-dimensional setting that achieves faster rates, suggesting that certain dimensional thresholds are intrinsic to the complexity of non-convex stochastic optimization.
- [659] arXiv:2502.17358 (replaced) [pdf, other]
-
Title: DIS-CO: Discovering Copyrighted Content in VLMs Training DataSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
How can we verify whether copyrighted content was used to train a large vision-language model (VLM) without direct access to its training data? Motivated by the hypothesis that a VLM is able to recognize images from its training corpus, we propose DIS-CO, a novel approach to infer the inclusion of copyrighted content during the model's development. By repeatedly querying a VLM with specific frames from targeted copyrighted material, DIS-CO extracts the content's identity through free-form text completions. To assess its effectiveness, we introduce MovieTection, a benchmark comprising 14,000 frames paired with detailed captions, drawn from films released both before and after a model's training cutoff. Our results show that DIS-CO significantly improves detection performance, nearly doubling the average AUC of the best prior method on models with logits available. Our findings also highlight a broader concern: all tested models appear to have been exposed to some extent to copyrighted content. Our code and data are available at this https URL
- [660] arXiv:2502.17701 (replaced) [pdf, html, other]
-
Title: From Perceptions to Decisions: Wildfire Evacuation Decision Prediction with Behavioral Theory-informed LLMsComments: 25 pages, 9 figuresSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computers and Society (cs.CY); Machine Learning (cs.LG)
Evacuation decision prediction is critical for efficient and effective wildfire response by helping emergency management anticipate traffic congestion and bottlenecks, allocate resources, and minimize negative impacts. Traditional statistical methods for evacuation decision prediction fail to capture the complex and diverse behavioral logic of different individuals. In this work, for the first time, we introduce FLARE, short for facilitating LLM for advanced reasoning on wildfire evacuation decision prediction, a Large Language Model (LLM)-based framework that integrates behavioral theories and models to streamline the Chain-of-Thought (CoT) reasoning and subsequently integrate with memory-based Reinforcement Learning (RL) module to provide accurate evacuation decision prediction and understanding. Our proposed method addresses the limitations of using existing LLMs for evacuation behavioral predictions, such as limited survey data, mismatching with behavioral theory, conflicting individual preferences, implicit and complex mental states, and intractable mental state-behavior mapping. Experiments on three post-wildfire survey datasets show an average of 20.47% performance improvement over traditional theory-informed behavioral models, with strong cross-event generalizability. Our complete code is publicly available at this https URL
- [661] arXiv:2502.18530 (replaced) [pdf, html, other]
-
Title: IMPROVE: Iterative Model Pipeline Refinement and Optimization Leveraging LLM ExpertsSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Large language model (LLM) agents have emerged as a promising solution to automate the workflow of machine learning, but most existing methods share a common limitation: they attempt to optimize entire pipelines in a single step before evaluation, making it difficult to attribute improvements to specific changes. This lack of granularity leads to unstable optimization and slower convergence, limiting their effectiveness. To address this, we introduce Iterative Refinement, a novel strategy for LLM-driven ML pipeline design inspired by how human ML experts iteratively refine models, focusing on one component at a time rather than making sweeping changes all at once. By systematically updating individual components based on real training feedback, Iterative Refinement improves overall model performance. We also provide some theoretical edvience of the superior properties of this Iterative Refinement. Further, we implement this strategy in IMPROVE, an end-to-end LLM agent framework for automating and optimizing object classification pipelines. Through extensive evaluations across datasets of varying sizes and domains, we demonstrate that Iterative Refinement enables IMPROVE to consistently achieve better performance over existing zero-shot LLM-based approaches.
- [662] arXiv:2502.19163 (replaced) [pdf, html, other]
-
Title: TestNUC: Enhancing Test-Time Computing Approaches and Scaling through Neighboring Unlabeled Data ConsistencyHenry Peng Zou, Zhengyao Gu, Yue Zhou, Yankai Chen, Weizhi Zhang, Liancheng Fang, Yibo Wang, Yangning Li, Kay Liu, Philip S. YuComments: Accepted by ACL 2025 main conferenceSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Machine Learning (cs.LG)
Test-time computing approaches, which leverage additional computational resources during inference, have been proven effective in enhancing large language model performance. This work introduces a novel, linearly scaling approach, TestNUC, that improves test-time predictions by leveraging the local consistency of neighboring unlabeled data-it classifies an input instance by considering not only the model's prediction on that instance but also on neighboring unlabeled instances. We evaluate TestNUC across eight diverse datasets, spanning intent classification, topic mining, domain discovery, and emotion detection, demonstrating its consistent superiority over baseline methods such as standard prompting and self-consistency. Furthermore, TestNUC can be seamlessly integrated with existing test-time computing approaches, substantially boosting their performance. Our analysis reveals that TestNUC scales effectively with increasing amounts of unlabeled data and performs robustly across different embedding models, making it practical for real-world applications. Our code is available at this https URL.
- [663] arXiv:2502.19765 (replaced) [pdf, html, other]
-
Title: EdiText: Controllable Coarse-to-Fine Text Editing with Diffusion Language ModelsComments: ACL 2025Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
We propose EdiText, a controllable text editing method that modifies the reference text to desired attributes at various scales. We integrate an SDEdit-based editing technique that allows for broad adjustments in the degree of text editing. Additionally, we introduce a novel fine-level editing method based on self-conditioning, which allows subtle control of reference text. While being capable of editing on its own, this fine-grained method, integrated with the SDEdit approach, enables EdiText to make precise adjustments within the desired range. EdiText demonstrates its controllability to robustly adjust reference text at a broad range of levels across various tasks, including toxicity control and sentiment control.
- [664] arXiv:2502.19952 (replaced) [pdf, html, other]
-
Title: Towards Collaborative Anti-Money Laundering Among Financial InstitutionsComments: Accepted by International World Wide Web Conference (WWW) 2025Subjects: Social and Information Networks (cs.SI); Computers and Society (cs.CY); Machine Learning (cs.LG)
Money laundering is the process that intends to legalize the income derived from illicit activities, thus facilitating their entry into the monetary flow of the economy without jeopardizing their source. It is crucial to identify such activities accurately and reliably in order to enforce anti-money laundering (AML). Despite considerable efforts to AML, a large number of such activities still go undetected. Rule-based methods were first introduced and are still widely used in current detection systems. With the rise of machine learning, graph-based learning methods have gained prominence in detecting illicit accounts through the analysis of money transfer graphs. Nevertheless, these methods generally assume that the transaction graph is centralized, whereas in practice, money laundering activities usually span multiple financial institutions. Due to regulatory, legal, commercial, and customer privacy concerns, institutions tend not to share data, restricting their utility in practical usage. In this paper, we propose the first algorithm that supports performing AML over multiple institutions while protecting the security and privacy of local data. To evaluate, we construct Alipay-ECB, a real-world dataset comprising digital transactions from Alipay, the world's largest mobile payment platform, alongside transactions from E-Commerce Bank (ECB). The dataset includes over 200 million accounts and 300 million transactions, covering both intra-institution transactions and those between Alipay and ECB. This makes it the largest real-world transaction graph available for analysis. The experimental results demonstrate that our methods can effectively identify cross-institution money laundering subgroups. Additionally, experiments on synthetic datasets also demonstrate that our method is efficient, requiring only a few minutes on datasets with millions of transactions.
- [665] arXiv:2502.20727 (replaced) [pdf, html, other]
-
Title: SPD: Sync-Point Drop for Efficient Tensor Parallelism of Large Language ModelsComments: International Conference on Machine Learning (ICML) 2025Subjects: Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
With the rapid expansion in the scale of large language models (LLMs), enabling efficient distributed inference across multiple computing units has become increasingly critical. However, communication overheads from popular distributed inference techniques such as Tensor Parallelism pose a significant challenge to achieve scalability and low latency. Therefore, we introduce a novel optimization technique, Sync-Point Drop (SPD), to reduce communication overheads in tensor parallelism by selectively dropping synchronization on attention outputs. In detail, we first propose a block design that allows execution to proceed without communication through SPD. Second, we apply different SPD strategies to attention blocks based on their sensitivity to the model accuracy. The proposed methods effectively alleviate communication bottlenecks while minimizing accuracy degradation during LLM inference, offering a scalable solution for diverse distributed environments: SPD offered about 20% overall inference latency reduction with < 1% accuracy regression for LLaMA2-70B inference over 8 GPUs.
- [666] arXiv:2503.05750 (replaced) [pdf, html, other]
-
Title: CSTRL: Context-Driven Sequential Transfer Learning for Abstractive Radiology Report SummarizationMst. Fahmida Sultana Naznin, Adnan Ibney Faruq, Mostafa Rifat Tazwar, Md Jobayer, Md. Mehedi Hasan Shawon, Md Rakibul HasanComments: Accepted in ACL 2025 FindingsSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
A radiology report comprises several sections, including the Findings and Impression of the diagnosis. Automatically generating the Impression from the Findings is crucial for reducing radiologists' workload and improving diagnostic accuracy. Pretrained models that excel in common abstractive summarization problems encounter challenges when applied to specialized medical domains largely due to the complex terminology and the necessity for accurate clinical context. Such tasks in medical domains demand extracting core information, avoiding context shifts, and maintaining proper flow. Misuse of medical terms can lead to drastic clinical errors. To address these issues, we introduce a sequential transfer learning that ensures key content extraction and coherent summarization. Sequential transfer learning often faces challenges like initial parameter decay and knowledge loss, which we resolve with the Fisher matrix regularization. Using MIMIC-CXR and Open-I datasets, our model, CSTRL - Context-driven Sequential TRansfer Learning - achieved state-of-the-art performance, showing 56.2% improvement in BLEU-1, 40.5% in BLEU-2, 84.3% in BLEU-3, 28.9% in ROUGE-1, 41.0% in ROUGE-2 and 26.5% in ROGUE-3 score over benchmark studies. We also analyze factual consistency scores while preserving the medical context. Our code is publicly available at this https URL.
- [667] arXiv:2503.05763 (replaced) [pdf, html, other]
-
Title: GMLM: Bridging Graph Neural Networks and Language Models for Heterophilic Node ClassificationSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Integrating structured graph data with rich textual information from nodes poses a significant challenge, particularly for heterophilic node classification. Current approaches often struggle with computational costs or effective fusion of disparate modalities. We propose \textbf{Graph Masked Language Model (GMLM)}, a novel architecture efficiently combining Graph Neural Networks (GNNs) with Pre-trained Language Models (PLMs). GMLM introduces three key innovations: (i) a \textbf{dynamic active node selection} strategy for scalable PLM text processing; (ii) a GNN-specific \textbf{contrastive pretraining stage} using soft masking with a learnable graph \texttt{[MASK]} token for robust structural representations; and (iii) a \textbf{dedicated fusion module} integrating RGCN-based GNN embeddings with PLM (GTE-Small \& DistilBERT) embeddings. Extensive experiments on heterophilic benchmarks (Cornell, Wisconsin, Texas) demonstrate GMLM's superiority. Notably, GMLM(DistilBERT) achieves significant performance gains, improving accuracy by over \textbf{4.7\%} on Cornell and over \textbf{2.0\%} on Texas compared to the previous best-performing baselines. This work underscores the benefits of targeted PLM engagement and modality-specific pretraining for improved, efficient learning on text-rich graphs.
- [668] arXiv:2503.15704 (replaced) [pdf, html, other]
-
Title: Tuning Sequential Monte Carlo Samplers via Greedy Incremental Divergence MinimizationComments: Accepted to ICML'25Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Computation (stat.CO)
The performance of sequential Monte Carlo (SMC) samplers heavily depends on the tuning of the Markov kernels used in the path proposal. For SMC samplers with unadjusted Markov kernels, standard tuning objectives, such as the Metropolis-Hastings acceptance rate or the expected-squared jump distance, are no longer applicable. While stochastic gradient-based end-to-end optimization has been explored for tuning SMC samplers, they often incur excessive training costs, even for tuning just the kernel step sizes. In this work, we propose a general adaptation framework for tuning the Markov kernels in SMC samplers by minimizing the incremental Kullback-Leibler (KL) divergence between the proposal and target paths. For step size tuning, we provide a gradient- and tuning-free algorithm that is generally applicable for kernels such as Langevin Monte Carlo (LMC). We further demonstrate the utility of our approach by providing a tailored scheme for tuning kinetic LMC used in SMC samplers. Our implementations are able to obtain a full schedule of tuned parameters at the cost of a few vanilla SMC runs, which is a fraction of gradient-based approaches.
- [669] arXiv:2503.16315 (replaced) [pdf, other]
-
Title: Active Learning For Repairable Hardware Systems With Partial CoverageComments: Submitted to IEEE Reliability and Maintainability Symposium - Europe 2025Subjects: Applications (stat.AP); Machine Learning (cs.LG)
Identifying the optimal diagnostic test and hardware system instance to infer reliability characteristics using field data is challenging, especially when constrained by fixed budgets and minimal maintenance cycles. Active Learning (AL) has shown promise for parameter inference with limited data and budget constraints in machine learning/deep learning tasks. However, AL for reliability model parameter inference remains underexplored for repairable hardware systems. It requires specialized AL Acquisition Functions (AFs) that consider hardware aging and the fact that a hardware system consists of multiple sub-systems, which may undergo only partial testing during a given diagnostic test. To address these challenges, we propose a relaxed Mixed Integer Semidefinite Program (MISDP) AL AF that incorporates Diagnostic Coverage (DC), Fisher Information Matrices (FIMs), and diagnostic testing budgets. Furthermore, we design empirical-based simulation experiments focusing on two diagnostic testing scenarios: (1) partial tests of a hardware system with overlapping subsystem coverage, and (2) partial tests where one diagnostic test fully subsumes the subsystem coverage of another. We evaluate our proposed approach against the most widely used AL AF in the literature (entropy), as well as several intuitive AL AFs tailored for reliability model parameter inference. Our proposed AF ranked best on average among the alternative AFs across 6,000 experimental configurations, with respect to Area Under the Curve (AUC) of the Absolute Total Expected Event Error (ATEER) and Mean Squared Error (MSE) curves, with statistical significance calculated at a 0.05 alpha level using a Friedman hypothesis test.
- [670] arXiv:2503.17414 (replaced) [pdf, html, other]
-
Title: Opportunities and Challenges of Frontier Data Governance With Synthetic DataComments: Published at the ICLR 2025 Workshop on Human-AI Coevolution (HAIC)Subjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Synthetic data, or data generated by machine learning models, is increasingly emerging as a solution to the data access problem. However, its use introduces significant governance and accountability challenges, and potentially debases existing governance paradigms, such as compute and data governance. In this paper, we identify 3 key governance and accountability challenges that synthetic data poses - it can enable the increased emergence of malicious actors, spontaneous biases and value drift. We thus craft 3 technical mechanisms to address these specific challenges, finding applications for synthetic data towards adversarial training, bias mitigation and value reinforcement. These could not only counteract the risks of synthetic data, but serve as critical levers for governance of the frontier in the future.
- [671] arXiv:2503.18938 (replaced) [pdf, html, other]
-
Title: AdaWorld: Learning Adaptable World Models with Latent ActionsSubjects: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Robotics (cs.RO)
World models aim to learn action-controlled future prediction and have proven essential for the development of intelligent agents. However, most existing world models rely heavily on substantial action-labeled data and costly training, making it challenging to adapt to novel environments with heterogeneous actions through limited interactions. This limitation can hinder their applicability across broader domains. To overcome this limitation, we propose AdaWorld, an innovative world model learning approach that enables efficient adaptation. The key idea is to incorporate action information during the pretraining of world models. This is achieved by extracting latent actions from videos in a self-supervised manner, capturing the most critical transitions between frames. We then develop an autoregressive world model that conditions on these latent actions. This learning paradigm enables highly adaptable world models, facilitating efficient transfer and learning of new actions even with limited interactions and finetuning. Our comprehensive experiments across multiple environments demonstrate that AdaWorld achieves superior performance in both simulation quality and visual planning.
- [672] arXiv:2503.19449 (replaced) [pdf, html, other]
-
Title: VecTrans: Enhancing Compiler Auto-Vectorization through LLM-Assisted Code TransformationsZhongchun Zheng, Kan Wu, Long Cheng, Lu Li, Rodrigo C. O. Rocha, Tianyi Liu, Wei Wei, Jianjiang Zeng, Xianwei Zhang, Yaoqing GaoSubjects: Software Engineering (cs.SE); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Performance (cs.PF)
Auto-vectorization is a fundamental optimization for modern compilers to exploit SIMD parallelism. However, state-of-the-art approaches still struggle to handle intricate code patterns, often requiring manual hints or domain-specific expertise. Large language models (LLMs), with their ability to capture intricate patterns, provide a promising solution, yet their effective application in compiler optimizations remains an open challenge due to issues such as hallucinations and a lack of domain-specific reasoning. In this paper, we present VecTrans, a novel framework that leverages LLMs to enhance compiler-based code vectorization. VecTrans first employs compiler analysis to identify potentially vectorizable code regions. It then utilizes an LLM to refactor these regions into patterns that are more amenable to the compilers auto-vectorization. To ensure semantic correctness, VecTrans further integrates a hybrid validation mechanism at the intermediate representation (IR) level. With the above efforts, VecTrans combines the adaptability of LLMs with the precision of compiler vectorization, thereby effectively opening up the vectorization opportunities. experimental results show that among all TSVC functions unvectorizable by GCC, ICC, Clang, and BiSheng Compiler, VecTrans achieves an geomean speedup of 1.77x and successfully vectorizes 24 of 51 test cases. This marks a significant advancement over state-of-the-art approaches while maintaining a cost efficiency of $0.012 per function optimization for LLM API usage.
- [673] arXiv:2503.20158 (replaced) [pdf, html, other]
-
Title: RxRx3-core: Benchmarking drug-target interactions in High-Content MicroscopyOren Kraus, Federico Comitani, John Urbanik, Kian Kenyon-Dean, Lakshmanan Arumugam, Saber Saberian, Cas Wognum, Safiye Celik, Imran S. HaqueComments: Published at LMRL Workshop at ICLR 2025; Updated gene-gene interaction results based on filtering compounds wells out before computing known_relationship_benchmarkSubjects: Quantitative Methods (q-bio.QM); Machine Learning (cs.LG); Cell Behavior (q-bio.CB)
High Content Screening (HCS) microscopy datasets have transformed the ability to profile cellular responses to genetic and chemical perturbations, enabling cell-based inference of drug-target interactions (DTI). However, the adoption of representation learning methods for HCS data has been hindered by the lack of accessible datasets and robust benchmarks. To address this gap, we present RxRx3-core, a curated and compressed subset of the RxRx3 dataset, and an associated DTI benchmarking task. At just 18GB, RxRx3-core significantly reduces the size barrier associated with large-scale HCS datasets while preserving critical data necessary for benchmarking representation learning models against a zero-shot DTI prediction task. RxRx3-core includes 222,601 microscopy images spanning 736 CRISPR knockouts and 1,674 compounds at 8 concentrations. RxRx3-core is available on HuggingFace and Polaris, along with pre-trained embeddings and benchmarking code, ensuring accessibility for the research community. By providing a compact dataset and robust benchmarks, we aim to accelerate innovation in representation learning methods for HCS data and support the discovery of novel biological insights.
- [674] arXiv:2503.20756 (replaced) [pdf, html, other]
-
Title: ADS-Edit: A Multimodal Knowledge Editing Dataset for Autonomous Driving SystemsComments: Work in progressSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Multimedia (cs.MM)
Recent advancements in Large Multimodal Models (LMMs) have shown promise in Autonomous Driving Systems (ADS). However, their direct application to ADS is hindered by challenges such as misunderstanding of traffic knowledge, complex road conditions, and diverse states of vehicle. To address these challenges, we propose the use of Knowledge Editing, which enables targeted modifications to a model's behavior without the need for full retraining. Meanwhile, we introduce ADS-Edit, a multimodal knowledge editing dataset specifically designed for ADS, which includes various real-world scenarios, multiple data types, and comprehensive evaluation metrics. We conduct comprehensive experiments and derive several interesting conclusions. We hope that our work will contribute to the further advancement of knowledge editing applications in the field of autonomous driving. Code and data are available in this https URL.
- [675] arXiv:2503.20787 (replaced) [pdf, html, other]
-
Title: Advanced simulation paradigm of human behaviour unveils complex financial systemic projectionSubjects: Trading and Market Microstructure (q-fin.TR); Machine Learning (cs.LG)
The high-order complexity of human behaviour is likely the root cause of extreme difficulty in financial market projections. We consider that behavioural simulation can unveil systemic dynamics to support analysis. Simulating diverse human groups must account for the behavioural heterogeneity, especially in finance. To address the fidelity of simulated agents, on the basis of agent-based modeling, we propose a new paradigm of behavioural simulation where each agent is supported and driven by a hierarchical knowledge architecture. This architecture, integrating language and professional models, imitates behavioural processes in specific scenarios. Evaluated on futures markets, our simulator achieves a 13.29% deviation in simulating crisis scenarios whose price increase rate reaches 285.34%. Under normal conditions, our simulator also exhibits lower mean square error in predicting futures price of specific commodities. This technique bridges non-quantitative information with diverse market behaviour, offering a promising platform to simulate investor behaviour and its impact on market dynamics.
- [676] arXiv:2503.23487 (replaced) [pdf, other]
-
Title: Large Language and Reasoning Models are Shallow Disjunctive ReasonersComments: ACL 2025 main conferenceSubjects: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Large Language Models (LLMs) have been found to struggle with systematic reasoning. Even on tasks where they appear to perform well, their performance often depends on shortcuts, rather than on genuine reasoning abilities, leading them to collapse on out-of-distribution (OOD) examples. Post-training strategies based on reinforcement learning and chain-of-thought prompting have recently been hailed as a step change. However, little is known about the potential of the resulting ``Large Reasoning Models'' (LRMs) beyond maths and programming-based problem solving, where genuine OOD problems can be sparse. In this paper, we focus on tasks that require systematic relational composition for qualitative spatial and temporal reasoning. The setting allows fine control over problem difficulty to precisely measure OOD generalization. We find that, zero-shot LRMs generally outperform their LLM counterparts in single-path reasoning tasks but struggle in the multi-path setting. Whilst showing comparatively better results, fine-tuned LLMs are also not capable of multi-path generalization. We also provide evidence for the behavioral interpretation for this, i.e., that LRMs are shallow disjunctive reasoners.
- [677] arXiv:2504.03561 (replaced) [pdf, html, other]
-
Title: SynWorld: Virtual Scenario Synthesis for Agentic Action Knowledge RefinementRunnan Fang, Xiaobin Wang, Yuan Liang, Shuofei Qiao, Jialong Wu, Zekun Xi, Ningyu Zhang, Yong Jiang, Pengjun Xie, Fei Huang, Huajun ChenComments: ACL 2025Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
In the interaction between agents and their environments, agents expand their capabilities by planning and executing actions. However, LLM-based agents face substantial challenges when deployed in novel environments or required to navigate unconventional action spaces. To empower agents to autonomously explore environments, optimize workflows, and enhance their understanding of actions, we propose SynWorld, a framework that allows agents to synthesize possible scenarios with multi-step action invocation within the action space and perform Monte Carlo Tree Search (MCTS) exploration to effectively refine their action knowledge in the current environment. Our experiments demonstrate that SynWorld is an effective and general approach to learning action knowledge in new environments. Code is available at this https URL.
- [678] arXiv:2504.05336 (replaced) [pdf, html, other]
-
Title: Quantum Adaptive Self-Attention for Quantum Transformer ModelsSubjects: Quantum Physics (quant-ph); Machine Learning (cs.LG)
Transformer models have revolutionized sequential learning across various domains, yet their self-attention mechanism incurs quadratic computational cost, posing limitations for real-time and resource-constrained tasks. To address this, we propose Quantum Adaptive Self-Attention (QASA), a novel hybrid architecture that enhances classical Transformer models with a quantum attention mechanism. QASA replaces dot-product attention with a parameterized quantum circuit (PQC) that adaptively captures inter-token relationships in the quantum Hilbert space. Additionally, a residual quantum projection module is introduced before the feedforward network to further refine temporal features. Our design retains classical efficiency in earlier layers while injecting quantum expressiveness in the final encoder block, ensuring compatibility with current NISQ hardware. Experiments on synthetic time-series tasks demonstrate that QASA achieves faster convergence and superior generalization compared to both standard Transformers and reduced classical variants. Preliminary complexity analysis suggests potential quantum advantages in gradient computation, opening new avenues for efficient quantum deep learning models.
- [679] arXiv:2504.16449 (replaced) [pdf, html, other]
-
Title: From Past to Present: A Survey of Malicious URL Detection Techniques, Datasets and Code RepositoriesComments: This manuscript is currently under review at Computer Science ReviewSubjects: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
Malicious URLs persistently threaten the cybersecurity ecosystem, by either deceiving users into divulging private data or distributing harmful payloads to infiltrate host systems. Gaining timely insights into the current state of this ongoing battle holds significant importance. However, existing reviews exhibit 4 critical gaps: 1) Their reliance on algorithm-centric taxonomies obscures understanding of how detection approaches exploit specific modal information channels; 2) They fail to incorporate pivotal LLM/Transformer-based defenses; 3) No open-source implementations are collected to facilitate benchmarking; 4) Insufficient dataset this http URL paper presents a comprehensive review of malicious URL detection technologies, systematically analyzing methods from traditional blacklisting to advanced deep learning approaches (e.g. Transformer, GNNs, and LLMs). Unlike prior surveys, we propose a novel modality-based taxonomy that categorizes existing works according to their primary data modalities (URL, HTML, Visual, etc.). This hierarchical classification enables both rigorous technical analysis and clear understanding of multimodal information utilization. Furthermore, to establish a profile of accessible datasets and address the lack of standardized benchmarking (where current studies often lack proper baseline comparisons), we curate and analyze: 1) publicly available datasets (2016-2024), and 2) open-source implementations from published works(2013-2025). Then, we outline essential design principles and architectural frameworks for product-level implementations. The review concludes by examining emerging challenges and proposing actionable directions for future research. We maintain a GitHub repository for ongoing curating datasets and open-source implementations: this https URL.
- [680] arXiv:2504.19475 (replaced) [pdf, html, other]
-
Title: Prisma: An Open Source Toolkit for Mechanistic Interpretability in Vision and VideoSonia Joseph, Praneet Suresh, Lorenz Hufe, Edward Stevinson, Robert Graham, Yash Vadi, Danilo Bzdok, Sebastian Lapuschkin, Lee Sharkey, Blake Aaron RichardsComments: 4 pages, 3 figures, 9 tables. Oral and Tutorial at the CVPR Mechanistic Interpretability for Vision (MIV) WorkshopSubjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Robust tooling and publicly available pre-trained models have helped drive recent advances in mechanistic interpretability for language models. However, similar progress in vision mechanistic interpretability has been hindered by the lack of accessible frameworks and pre-trained weights. We present Prisma (Access the codebase here: this https URL), an open-source framework designed to accelerate vision mechanistic interpretability research, providing a unified toolkit for accessing 75+ vision and video transformers; support for sparse autoencoder (SAE), transcoder, and crosscoder training; a suite of 80+ pre-trained SAE weights; activation caching, circuit analysis tools, and visualization tools; and educational resources. Our analysis reveals surprising findings, including that effective vision SAEs can exhibit substantially lower sparsity patterns than language SAEs, and that in some instances, SAE reconstructions can decrease model loss. Prisma enables new research directions for understanding vision model internals while lowering barriers to entry in this emerging field.
- [681] arXiv:2505.00763 (replaced) [pdf, html, other]
-
Title: JFlow: Model-Independent Spherical Jeans Analysis using Equivariant Continuous Normalizing FlowsComments: 10 pages, 3 figures, 1 table, revised version for the journal submissionSubjects: Astrophysics of Galaxies (astro-ph.GA); Cosmology and Nongalactic Astrophysics (astro-ph.CO); Machine Learning (cs.LG); High Energy Physics - Experiment (hep-ex); High Energy Physics - Phenomenology (hep-ph)
The kinematics of stars in dwarf spheroidal galaxies have been studied to understand the structure of dark matter halos. However, the kinematic information of these stars is often limited to celestial positions and line-of-sight velocities, making full phase space analysis challenging. Conventional methods rely on projected analytic phase space density models with several parameters and infer dark matter halo structures by solving the spherical Jeans equation. In this paper, we introduce an unsupervised machine learning method for solving the spherical Jeans equation in a model-independent way as a first step toward model-independent analysis of dwarf spheroidal galaxies. Using equivariant continuous normalizing flows, we demonstrate that spherically symmetric stellar phase space densities and velocity dispersions can be estimated without model assumptions. As a proof of concept, we apply our method to Gaia challenge datasets for spherical models and measure dark matter mass densities for given velocity anisotropy profiles. Our method can identify halo structures accurately, even with a small number of tracer stars.
- [682] arXiv:2505.01136 (replaced) [pdf, html, other]
-
Title: Descriptor: C++ Self-Admitted Technical Debt Dataset (CppSATD)Subjects: Software Engineering (cs.SE); Information Retrieval (cs.IR); Machine Learning (cs.LG); Programming Languages (cs.PL)
In software development, technical debt (TD) refers to suboptimal implementation choices made by the developers to meet urgent deadlines and limited resources, posing challenges for future maintenance. Self-Admitted Technical Debt (SATD) is a sub-type of TD, representing specific TD instances ``openly admitted'' by the developers and often expressed through source code comments. Previous research on SATD has focused predominantly on the Java programming language, revealing a significant gap in cross-language SATD. Such a narrow focus limits the generalizability of existing findings as well as SATD detection techniques across multiple programming languages. Our work addresses such limitation by introducing CppSATD, a dedicated C++ SATD dataset, comprising over 531,000 annotated comments and their source code contexts. Our dataset can serve as a foundation for future studies that aim to develop SATD detection methods in C++, generalize the existing findings to other languages, or contribute novel insights to cross-language SATD research.
- [683] arXiv:2505.03201 (replaced) [pdf, html, other]
-
Title: Weighted Integrated Gradients for Feature AttributionSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG)
In explainable AI, Integrated Gradients (IG) is a widely adopted technique for assessing the significance of feature attributes of the input on model outputs by evaluating contributions from a baseline input to the current input. The choice of the baseline input significantly influences the resulting explanation. While the traditional Expected Gradients (EG) method assumes baselines can be uniformly sampled and averaged with equal weights, this study argues that baselines should not be treated equivalently. We introduce Weighted Integrated Gradients (WG), a novel approach that unsupervisedly evaluates baseline suitability and incorporates a strategy for selecting effective baselines. Theoretical analysis demonstrates that WG satisfies essential explanation method criteria and offers greater stability than prior approaches. Experimental results further confirm that WG outperforms EG across diverse scenarios, achieving an improvement of 10-35\% on main metrics. Moreover, by evaluating baselines, our method can filter a subset of effective baselines for each input to calculate explanations, maintaining high accuracy while reducing computational cost. The code is available at: this https URL.
- [684] arXiv:2505.04792 (replaced) [pdf, html, other]
-
Title: Confabulation dynamics in a reservoir computer: Filling in the gaps with untrained attractorsSubjects: Dynamical Systems (math.DS); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Artificial Intelligence has advanced significantly in recent years thanks to innovations in the design and training of artificial neural networks (ANNs). Despite these advancements, we still understand relatively little about how elementary forms of ANNs learn, fail to learn, and generate false information without the intent to deceive, a phenomenon known as `confabulation'. To provide some foundational insight, in this paper we analyse how confabulation occurs in reservoir computers (RCs): a dynamical system in the form of an ANN. RCs are particularly useful to study as they are known to confabulate in a well-defined way: when RCs are trained to reconstruct the dynamics of a given attractor, they sometimes construct an attractor that they were not trained to construct, a so-called `untrained attractor' (UA). This paper sheds light on the role played by UAs when reconstruction fails and their influence when modelling transitions between reconstructed attractors. Based on our results, we conclude that UAs are an intrinsic feature of learning systems whose state spaces are bounded, and that this means of confabulation may be present in systems beyond RCs.
- [685] arXiv:2505.06146 (replaced) [pdf, html, other]
-
Title: Learning-Augmented Algorithms for Boolean SatisfiabilitySubjects: Data Structures and Algorithms (cs.DS); Computational Complexity (cs.CC); Machine Learning (cs.LG)
Learning-augmented algorithms are a prominent recent development in beyond worst-case analysis. In this framework, a problem instance is provided with a prediction (``advice'') from a machine-learning oracle, which provides partial information about an optimal solution, and the goal is to design algorithms that leverage this advice to improve worst-case performance. We study the classic Boolean satisfiability (SAT) decision and optimization problems within this framework using two forms of advice. ``Subset advice" provides a random $\epsilon$ fraction of the variables from an optimal assignment, whereas ``label advice" provides noisy predictions for all variables in an optimal assignment.
For the decision problem $k$-SAT, by using the subset advice we accelerate the exponential running time of the PPSZ family of algorithms due to Paturi, Pudlak, Saks and Zane, which currently represent the state of the art in the worst case. We accelerate the running time by a multiplicative factor of $2^{-c}$ in the base of the exponent, where $c$ is a function of $\epsilon$ and $k$. For the optimization problem, we show how to incorporate subset advice in a black-box fashion with any $\alpha$-approximation algorithm, improving the approximation ratio to $\alpha + (1 - \alpha)\epsilon$. Specifically, we achieve approximations of $0.94 + \Omega(\epsilon)$ for MAX-$2$-SAT, $7/8 + \Omega(\epsilon)$ for MAX-$3$-SAT, and $0.79 + \Omega(\epsilon)$ for MAX-SAT. Moreover, for label advice, we obtain near-optimal approximation for instances with large average degree, thereby generalizing recent results on MAX-CUT and MAX-$2$-LIN. - [686] arXiv:2505.10573 (replaced) [pdf, html, other]
-
Title: Measurement to Meaning: A Validity-Centered Framework for AI EvaluationOlawale Salaudeen, Anka Reuel, Ahmed Ahmed, Suhana Bedi, Zachary Robertson, Sudharsan Sundar, Ben Domingue, Angelina Wang, Sanmi KoyejoComments: Correspondence to olawale@mit.eduSubjects: Computers and Society (cs.CY); Machine Learning (cs.LG)
While the capabilities and utility of AI systems have advanced, rigorous norms for evaluating these systems have lagged. Grand claims, such as models achieving general reasoning capabilities, are supported with model performance on narrow benchmarks, like performance on graduate-level exam questions, which provide a limited and potentially misleading assessment. We provide a structured approach for reasoning about the types of evaluative claims that can be made given the available evidence. For instance, our framework helps determine whether performance on a mathematical benchmark is an indication of the ability to solve problems on math tests or instead indicates a broader ability to reason. Our framework is well-suited for the contemporary paradigm in machine learning, where various stakeholders provide measurements and evaluations that downstream users use to validate their claims and decisions. At the same time, our framework also informs the construction of evaluations designed to speak to the validity of the relevant claims. By leveraging psychometrics' breakdown of validity, evaluations can prioritize the most critical facets for a given claim, improving empirical utility and decision-making efficacy. We illustrate our framework through detailed case studies of vision and language model evaluations, highlighting how explicitly considering validity strengthens the connection between evaluation evidence and the claims being made.
- [687] arXiv:2505.12185 (replaced) [pdf, html, other]
-
Title: EVALOOP: Assessing LLM Robustness in Programming from a Self-consistency PerspectiveComments: 19 pages, 11 figuresSubjects: Software Engineering (cs.SE); Computation and Language (cs.CL); Machine Learning (cs.LG)
Assessing the programming capabilities of Large Language Models (LLMs) is crucial for their effective use in software engineering. Current evaluations, however, predominantly measure the accuracy of generated code on static benchmarks, neglecting the critical aspect of model robustness during programming tasks. While adversarial attacks offer insights on model robustness, their effectiveness is limited and evaluation could be constrained. Current adversarial attack methods for robustness evaluation yield inconsistent results, struggling to provide a unified evaluation across different LLMs. We introduce EVALOOP, a novel assessment framework that evaluate the robustness from a self-consistency perspective, i.e., leveraging the natural duality inherent in popular software engineering tasks, e.g., code generation and code summarization. EVALOOP initiates a self-contained feedback loop: an LLM generates output (e.g., code) from an input (e.g., natural language specification), and then use the generated output as the input to produce a new output (e.g., summarizes that code into a new specification). EVALOOP repeats the process to assess the effectiveness of EVALOOP in each loop. This cyclical strategy intrinsically evaluates robustness without rely on any external attack setups, providing a unified metric to evaluate LLMs' robustness in programming. We evaluate 16 prominent LLMs (e.g., GPT-4.1, O4-mini) on EVALOOP and found that EVALOOP typically induces a 5.01%-19.31% absolute drop in pass@1 performance within ten loops. Intriguingly, robustness does not always align with initial performance (i.e., one-time query); for instance, GPT-3.5-Turbo, despite superior initial code generation compared to DeepSeek-V2, demonstrated lower robustness over repeated evaluation loop.
- [688] arXiv:2505.12942 (replaced) [pdf, html, other]
-
Title: A3 : an Analytical Low-Rank Approximation Framework for AttentionJeffrey T. H. Wong, Cheng Zhang, Xinye Cao, Pedro Gimenes, George A. Constantinides, Wayne Luk, Yiren ZhaoSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Large language models have demonstrated remarkable performance; however, their massive parameter counts make deployment highly expensive. Low-rank approximation offers a promising compression solution, yet existing approaches have two main limitations: (1) They focus on minimizing the output error of individual linear layers, without considering the architectural characteristics of Transformers, and (2) they decompose a large weight matrix into two small low-rank matrices. Consequently, these methods often fall short compared to other compression techniques like pruning and quantization, and introduce runtime overhead such as the extra GEMM kernel launches for decomposed small matrices. To address these limitations, we propose $\tt A^\tt 3$, a post-training low-rank approximation framework. $\tt A^\tt 3$ splits a Transformer layer into three functional components, namely $\tt QK$, $\tt OV$, and $\tt MLP$. For each component, $\tt A^\tt 3$ provides an analytical solution that reduces the hidden dimension size inside each component while minimizing the component's functional loss ($\it i.e.$, error in attention scores, attention outputs, and MLP outputs). This approach directly reduces model sizes, KV cache sizes, and FLOPs without introducing any runtime overheads. In addition, it provides a new narrative in advancing the optimization problem from singular linear layer loss optimization toward improved end-to-end performance. Through extensive experiments, we show that $\tt A^\tt 3$ maintains superior performance compared to SoTAs. For example, under the same reduction budget in computation and memory, our low-rank approximated LLaMA 3.1-70B achieves a perplexity of 4.69 on WikiText-2, outperforming the previous SoTA's 7.87 by 3.18. We also demonstrate the versatility of $\tt A^\tt 3$, including KV cache compression, quantization, and mixed-rank assignments for enhanced performance.
- [689] arXiv:2505.15791 (replaced) [pdf, html, other]
-
Title: VARD: Efficient and Dense Fine-Tuning for Diffusion Models with Value-based RLComments: Under reviewSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Diffusion models have emerged as powerful generative tools across various domains, yet tailoring pre-trained models to exhibit specific desirable properties remains challenging. While reinforcement learning (RL) offers a promising solution,current methods struggle to simultaneously achieve stable, efficient fine-tuning and support non-differentiable rewards. Furthermore, their reliance on sparse rewards provides inadequate supervision during intermediate steps, often resulting in suboptimal generation quality. To address these limitations, dense and differentiable signals are required throughout the diffusion process. Hence, we propose VAlue-based Reinforced Diffusion (VARD): a novel approach that first learns a value function predicting expection of rewards from intermediate states, and subsequently uses this value function with KL regularization to provide dense supervision throughout the generation process. Our method maintains proximity to the pretrained model while enabling effective and stable training via backpropagation. Experimental results demonstrate that our approach facilitates better trajectory guidance, improves training efficiency and extends the applicability of RL to diffusion models optimized for complex, non-differentiable reward functions.
- [690] arXiv:2505.16223 (replaced) [pdf, html, other]
-
Title: MADCluster: Model-agnostic Anomaly Detection with Self-supervised Clustering NetworkComments: 24 pages, 9 figuresSubjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
In this paper, we propose MADCluster, a novel model-agnostic anomaly detection framework utilizing self-supervised clustering. MADCluster is applicable to various deep learning architectures and addresses the 'hypersphere collapse' problem inherent in existing deep learning-based anomaly detection methods. The core idea is to cluster normal pattern data into a 'single cluster' while simultaneously learning the cluster center and mapping data close to this center. Also, to improve expressiveness and enable effective single clustering, we propose a new 'One-directed Adaptive loss'. The optimization of this loss is mathematically proven. MADCluster consists of three main components: Base Embedder capturing high-dimensional temporal dynamics, Cluster Distance Mapping, and Sequence-wise Clustering for continuous center updates. Its model-agnostic characteristics are achieved by applying various architectures to the Base Embedder. Experiments on four time series benchmark datasets demonstrate that applying MADCluster improves the overall performance of comparative models. In conclusion, the compatibility of MADCluster shows potential for enhancing model performance across various architectures.
- [691] arXiv:2505.16415 (replaced) [pdf, html, other]
-
Title: Attributing Response to Context: A Jensen-Shannon Divergence Driven Mechanistic Study of Context Attribution in Retrieval-Augmented GenerationComments: Work in processSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Retrieval-Augmented Generation (RAG) leverages large language models (LLMs) combined with external contexts to enhance the accuracy and reliability of generated responses. However, reliably attributing generated content to specific context segments, context attribution, remains challenging due to the computationally intensive nature of current methods, which often require extensive fine-tuning or human annotation. In this work, we introduce a novel Jensen-Shannon Divergence driven method to Attribute Response to Context (ARC-JSD), enabling efficient and accurate identification of essential context sentences without additional fine-tuning or surrogate modelling. Evaluations on a wide range of RAG benchmarks, such as TyDi QA, Hotpot QA, and Musique, using instruction-tuned LLMs in different scales demonstrate superior accuracy and significant computational efficiency improvements compared to the previous surrogate-based method. Furthermore, our mechanistic analysis reveals specific attention heads and multilayer perceptron (MLP) layers responsible for context attribution, providing valuable insights into the internal workings of RAG models. Our code is available at this https URL
- [692] arXiv:2505.16900 (replaced) [pdf, html, other]
-
Title: Power-Law Decay Loss for Large Language Model Finetuning: A Theory PerspectiveComments: We are withdrawing this submission as the underlying experiment is currently incomplete. We require additional time to gather more data and supplement the existing findings to ensure a comprehensive and robust presentation. We intend to resubmit once these additions are finalizedSubjects: Computation and Language (cs.CL); Machine Learning (cs.LG)
During the finetuning stage of text generation tasks, standard cross-entropy loss treats all tokens equally. This can lead models to overemphasize high-frequency, low-information tokens, neglecting lower-frequency tokens crucial for specificity and informativeness in generated content. This paper introduces a novel loss function, Power-Law Decay Loss (PDL), specifically designed to optimize the finetuning process for text generation. The core motivation for PDL stems from observations in information theory and linguistics: the informativeness of a token is often inversely proportional to its frequency of occurrence. PDL re-weights the contribution of each token in the standard cross-entropy loss based on its frequency in the training corpus, following a power-law decay. Specifically, the weights for high-frequency tokens are reduced, while low-frequency, information-dense tokens are assigned higher weights. This mechanism guides the model during finetuning to focus more on learning and generating tokens that convey specific and unique information, thereby enhancing the quality, diversity, and informativeness of the generated text. We theoretically elaborate on the motivation and construction of PDL and discuss its potential applications and advantages across various text generation finetuning tasks, such as abstractive summarization, dialogue systems, and style transfer.
- [693] arXiv:2505.18174 (replaced) [pdf, html, other]
-
Title: NMCSE: Noise-Robust Multi-Modal Coupling Signal Estimation Method via Optimal Transport for Cardiovascular Disease DetectionSubjects: Signal Processing (eess.SP); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Electrocardiogram (ECG) and Phonocardiogram (PCG) signals are linked by a latent coupling signal representing the electrical-to-mechanical cardiac transformation. While valuable for cardiovascular disease (CVD) detection, this coupling signal is traditionally estimated using deconvolution methods that amplify noise, limiting clinical utility. In this paper, we propose Noise-Robust Multi-Modal Coupling Signal Estimation (NMCSE), which reformulates the problem as distribution matching via optimal transport theory. By jointly optimizing amplitude and temporal alignment, NMCSE mitigates noise amplification without additional preprocessing. Integrated with our Temporal-Spatial Feature Extraction network, NMCSE enables robust multi-modal CVD detection. Experiments on the PhysioNet 2016 dataset with realistic hospital noise demonstrate that NMCSE reduces estimation errors by approximately 30% in Mean Squared Error while maintaining higher Pearson Correlation Coefficients across all tested signal-to-noise ratios. Our approach achieves 97.38% accuracy and 0.98 AUC in CVD detection, outperforming state-of-the-art methods and demonstrating robust performance for real-world clinical applications.
- [694] arXiv:2505.18361 (replaced) [pdf, html, other]
-
Title: Task-Optimized Convolutional Recurrent Networks Align with Tactile Processing in the Rodent BrainComments: 9 pages, 8 figures, 5 tablesSubjects: Neurons and Cognition (q-bio.NC); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Robotics (cs.RO)
Tactile sensing remains far less understood in neuroscience and less effective in artificial systems compared to more mature modalities such as vision and language. We bridge these gaps by introducing a novel Encoder-Attender-Decoder (EAD) framework to systematically explore the space of task-optimized temporal neural networks trained on realistic tactile input sequences from a customized rodent whisker-array simulator. We identify convolutional recurrent neural networks (ConvRNNs) as superior encoders to purely feedforward and state-space architectures for tactile categorization. Crucially, these ConvRNN-encoder-based EAD models achieve neural representations closely matching rodent somatosensory cortex, saturating the explainable neural variability and revealing a clear linear relationship between supervised categorization performance and neural alignment. Furthermore, contrastive self-supervised ConvRNN-encoder-based EADs, trained with tactile-specific augmentations, match supervised neural fits, serving as an ethologically-relevant, label-free proxy.
For neuroscience, our findings highlight nonlinear recurrent processing as important for general-purpose tactile representations in somatosensory cortex, providing the first quantitative characterization of the underlying inductive biases in this system. For embodied AI, our results emphasize the importance of recurrent EAD architectures to handle realistic tactile inputs, along with tailored self-supervised learning methods for achieving robust tactile perception with the same type of sensors animals use to sense in unstructured environments. - [695] arXiv:2505.18458 (replaced) [pdf, html, other]
-
Title: A Survey of LLM $\times$ DATAXuanhe Zhou, Junxuan He, Wei Zhou, Haodong Chen, Zirui Tang, Haoyu Zhao, Xin Tong, Guoliang Li, Youmin Chen, Jun Zhou, Zhaojun Sun, Binyuan Hui, Shuo Wang, Conghui He, Zhiyuan Liu, Jingren Zhou, Fan WuComments: Please refer to the paper list at: this https URLSubjects: Databases (cs.DB); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
The integration of large language model (LLM) and data management (DATA) is rapidly redefining both domains. In this survey, we comprehensively review the bidirectional relationships. On the one hand, DATA4LLM, spanning large-scale data processing, storage, and serving, feeds LLMs with high quality, diversity, and timeliness of data required for stages like pre-training, post-training, retrieval-augmented generation, and agentic workflows: (i) Data processing for LLMs includes scalable acquisition, deduplication, filtering, selection, domain mixing, and synthetic augmentation; (ii) Data Storage for LLMs focuses on efficient data and model formats, distributed and heterogeneous storage hierarchies, KV-cache management, and fault-tolerant checkpointing; (iii) Data serving for LLMs tackles challenges in RAG (e.g., knowledge post-processing), LLM inference (e.g., prompt compression, data provenance), and training strategies (e.g., data packing and shuffling). On the other hand, in LLM4DATA, LLMs are emerging as general-purpose engines for data management. We review recent advances in (i) data manipulation, including automatic data cleaning, integration, discovery; (ii) data analysis, covering reasoning over structured, semi-structured, and unstructured data, and (iii) system optimization (e.g., configuration tuning, query rewriting, anomaly diagnosis), powered by LLM techniques like retrieval-augmented prompting, task-specialized fine-tuning, and multi-agent collaboration.
- [696] arXiv:2505.18918 (replaced) [pdf, html, other]
-
Title: ALPCAHUS: Subspace Clustering for Heteroscedastic DataComments: Manuscript submitted to IEEE Transactions on Signal Processing (TSP) pending reviewSubjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Signal Processing (eess.SP)
Principal component analysis (PCA) is a key tool in the field of data dimensionality reduction. Various methods have been proposed to extend PCA to the union of subspace (UoS) setting for clustering data that come from multiple subspaces like K-Subspaces (KSS). However, some applications involve heterogeneous data that vary in quality due to noise characteristics associated with each data sample. Heteroscedastic methods aim to deal with such mixed data quality. This paper develops a heteroscedastic-focused subspace clustering method, named ALPCAHUS, that can estimate the sample-wise noise variances and use this information to improve the estimate of the subspace bases associated with the low-rank structure of the data. This clustering algorithm builds on K-Subspaces (KSS) principles by extending the recently proposed heteroscedastic PCA method, named LR-ALPCAH, for clusters with heteroscedastic noise in the UoS setting. Simulations and real-data experiments show the effectiveness of accounting for data heteroscedasticity compared to existing clustering algorithms. Code available at this https URL.
- [697] arXiv:2505.19110 (replaced) [pdf, html, other]
-
Title: An Interpretable Representation Learning Approach for Diffusion Tensor ImagingVishwa Mohan Singh, Alberto Gaston Villagran Asiares, Luisa Sophie Schuhmacher, Kate Rendall, Simon Weißbrod, David Rügamer, Inga KörteComments: Accepted for publication at MIDL 2025Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Diffusion Tensor Imaging (DTI) tractography offers detailed insights into the structural connectivity of the brain, but presents challenges in effective representation and interpretation in deep learning models. In this work, we propose a novel 2D representation of DTI tractography that encodes tract-level fractional anisotropy (FA) values into a 9x9 grayscale image. This representation is processed through a Beta-Total Correlation Variational Autoencoder with a Spatial Broadcast Decoder to learn a disentangled and interpretable latent embedding. We evaluate the quality of this embedding using supervised and unsupervised representation learning strategies, including auxiliary classification, triplet loss, and SimCLR-based contrastive learning. Compared to the 1D Group deep neural network (DNN) baselines, our approach improves the F1 score in a downstream sex classification task by 15.74% and shows a better disentanglement than the 3D representation.
- [698] arXiv:2505.20529 (replaced) [pdf, html, other]
-
Title: Training Articulatory Inversion Models for Inter-Speaker ConsistencySubjects: Sound (cs.SD); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
Acoustic-to-Articulatory Inversion (AAI) attempts to model the inverse mapping from speech to articulation. Exact articulatory prediction from speech alone may be impossible, as speakers can choose different forms of articulation seemingly without reference to their vocal tract structure. However, once a speaker has selected an articulatory form, their productions vary minimally. Recent works in AAI have proposed adapting Self-Supervised Learning (SSL) models to single-speaker datasets, claiming that these single-speaker models provide a universal articulatory template. In this paper, we investigate whether SSL-adapted models trained on single and multi-speaker data produce articulatory targets which are consistent across speaker identities for English and Russian. We do this through the use of a novel evaluation method which extracts articulatory targets using minimal pair sets. We also present a training method which can improve interspeaker consistency using only speech data.
- [699] arXiv:2505.22232 (replaced) [pdf, html, other]
-
Title: Judging Quality Across Languages: A Multilingual Approach to Pretraining Data Filtering with Language ModelsMehdi Ali, Manuel Brack, Max Lübbering, Elias Wendt, Abbas Goher Khan, Richard Rutmann, Alex Jude, Maurice Kraus, Alexander Arno Weber, David Kaczér, Florian Mai, Lucie Flek, Rafet Sifa, Nicolas Flores-Herr, Joachim Köhler, Patrick Schramowski, Michael Fromm, Kristian KerstingComments: Project page available at this https URLSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
High-quality multilingual training data is essential for effectively pretraining large language models (LLMs). Yet, the availability of suitable open-source multilingual datasets remains limited. Existing state-of-the-art datasets mostly rely on heuristic filtering methods, restricting both their cross-lingual transferability and scalability. Here, we introduce JQL, a systematic approach that efficiently curates diverse and high-quality multilingual data at scale while significantly reducing computational demands. JQL distills LLMs' annotation capabilities into lightweight annotators based on pretrained multilingual embeddings. These models exhibit robust multilingual and cross-lingual performance, even for languages and scripts unseen during training. Evaluated empirically across 35 languages, the resulting annotation pipeline substantially outperforms current heuristic filtering methods like Fineweb2. JQL notably enhances downstream model training quality and increases data retention rates. Our research provides practical insights and valuable resources for multilingual data curation, raising the standards of multilingual dataset development.
- [700] arXiv:2505.22238 (replaced) [pdf, html, other]
-
Title: Yambda-5B -- A Large-Scale Multi-modal Dataset for Ranking And RetrievalA. Ploshkin, V. Tytskiy, A. Pismenny, V. Baikalov, E. Taychinov, A. Permiakov, D. Burlakov, E. Krofto, N. SavushkinSubjects: Information Retrieval (cs.IR); Machine Learning (cs.LG)
We present Yambda-5B, a large-scale open dataset sourced from the Yandex Music streaming platform. Yambda-5B contains 4.79 billion user-item interactions from 1 million users across 9.39 million tracks. The dataset includes two primary types of interactions: implicit feedback (listening events) and explicit feedback (likes, dislikes, unlikes and undislikes). In addition, we provide audio embeddings for most tracks, generated by a convolutional neural network trained on audio spectrograms. A key distinguishing feature of Yambda-5B is the inclusion of the is_organic flag, which separates organic user actions from recommendation-driven events. This distinction is critical for developing and evaluating machine learning algorithms, as Yandex Music relies on recommender systems to personalize track selection for users. To support rigorous benchmarking, we introduce an evaluation protocol based on a Global Temporal Split, allowing recommendation algorithms to be assessed in conditions that closely mirror real-world use. We report benchmark results for standard baselines (ItemKNN, iALS) and advanced models (SANSA, SASRec) using a variety of evaluation metrics. By releasing Yambda-5B to the community, we aim to provide a readily accessible, industrial-scale resource to advance research, foster innovation, and promote reproducible results in recommender systems.
- [701] arXiv:2505.22642 (replaced) [pdf, html, other]
-
Title: FastTD3: Simple, Fast, and Capable Reinforcement Learning for Humanoid ControlComments: Project webpage: this https URLSubjects: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Reinforcement learning (RL) has driven significant progress in robotics, but its complexity and long training times remain major bottlenecks. In this report, we introduce FastTD3, a simple, fast, and capable RL algorithm that significantly speeds up training for humanoid robots in popular suites such as HumanoidBench, IsaacLab, and MuJoCo Playground. Our recipe is remarkably simple: we train an off-policy TD3 agent with several modifications -- parallel simulation, large-batch updates, a distributional critic, and carefully tuned hyperparameters. FastTD3 solves a range of HumanoidBench tasks in under 3 hours on a single A100 GPU, while remaining stable during training. We also provide a lightweight and easy-to-use implementation of FastTD3 to accelerate RL research in robotics.
- [702] arXiv:2505.23436 (replaced) [pdf, other]
-
Title: Emergent Risk Awareness in Rational Agents under Resource ConstraintsDaniel Jarne Ornia, Nicholas Bishop, Joel Dyer, Wei-Chen Lee, Ani Calinescu, Doyne Farmer, Michael WooldridgeSubjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Advanced reasoning models with agentic capabilities (AI agents) are deployed to interact with humans and to solve sequential decision-making problems under (approximate) utility functions and internal models. When such problems have resource or failure constraints where action sequences may be forcibly terminated once resources are exhausted, agents face implicit trade-offs that reshape their utility-driven (rational) behaviour. Additionally, since these agents are typically commissioned by a human principal to act on their behalf, asymmetries in constraint exposure can give rise to previously unanticipated misalignment between human objectives and agent incentives. We formalise this setting through a survival bandit framework, provide theoretical and empirical results that quantify the impact of survival-driven preference shifts, identify conditions under which misalignment emerges and propose mechanisms to mitigate the emergence of risk-seeking or risk-averse behaviours. As a result, this work aims to increase understanding and interpretability of emergent behaviours of AI agents operating under such survival pressure, and offer guidelines for safely deploying such AI systems in critical resource-limited environments.
- [703] arXiv:2505.23657 (replaced) [pdf, html, other]
-
Title: Active Layer-Contrastive Decoding Reduces Hallucination in Large Language Model GenerationSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
Recent decoding methods improve the factuality of large language models (LLMs) by refining how the next token is selected during generation. These methods typically operate at the token level, leveraging internal representations to suppress superficial patterns. Nevertheless, LLMs remain prone to hallucinations, especially over longer contexts. In this paper, we propose Active Layer-Contrastive Decoding (ActLCD), a novel decoding strategy that actively decides when to apply contrasting layers during generation. By casting decoding as a sequential decision-making problem, ActLCD employs a reinforcement learning policy guided by a reward-aware classifier to optimize factuality beyond the token level. Our experiments demonstrate that ActLCD surpasses state-of-the-art methods across five benchmarks, showcasing its effectiveness in mitigating hallucinations in diverse generation scenarios.
- [704] arXiv:2505.23671 (replaced) [pdf, other]
-
Title: GSO: Challenging Software Optimization Tasks for Evaluating SWE-AgentsComments: Website: this https URLSubjects: Software Engineering (cs.SE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
Developing high-performance software is a complex task that requires specialized expertise. We introduce GSO, a benchmark for evaluating language models' capabilities in developing high-performance software. We develop an automated pipeline that generates and executes performance tests to analyze repository commit histories to identify 102 challenging optimization tasks across 10 codebases, spanning diverse domains and programming languages. An agent is provided with a codebase and performance test as a precise specification, and tasked to improve the runtime efficiency, which is measured against the expert developer optimization. Our quantitative evaluation reveals that leading SWE-Agents struggle significantly, achieving less than 5% success rate, with limited improvements even with inference-time scaling. Our qualitative analysis identifies key failure modes, including difficulties with low-level languages, practicing lazy optimization strategies, and challenges in accurately localizing bottlenecks. We release the code and artifacts of our benchmark along with agent trajectories to enable future research.
- [705] arXiv:2505.23786 (replaced) [pdf, other]
-
Title: Mind the Gap: A Practical Attack on GGUF QuantizationComments: ICML 2025Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
With the increasing size of frontier LLMs, post-training quantization has become the standard for memory-efficient deployment. Recent work has shown that basic rounding-based quantization schemes pose security risks, as they can be exploited to inject malicious behaviors into quantized models that remain hidden in full precision. However, existing attacks cannot be applied to more complex quantization methods, such as the GGUF family used in the popular `ollama` and `this http URL` frameworks. In this work, we address this gap by introducing the first attack on GGUF. Our key insight is that the quantization error -- the difference between the full-precision weights and their (de-)quantized version -- provides sufficient flexibility to construct malicious quantized models that appear benign in full precision. Leveraging this, we develop an attack that trains the target malicious LLM while constraining its weights based on quantization errors. We demonstrate the effectiveness of our attack on three popular LLMs across nine GGUF quantization data types on three diverse attack scenarios: insecure code generation ($\Delta$=$88.7\%$), targeted content injection ($\Delta$=$85.0\%$), and benign instruction refusal ($\Delta$=$30.1\%$). Our attack highlights that (1) the most widely used post-training quantization method is susceptible to adversarial interferences, and (2) the complexity of quantization schemes alone is insufficient as a defense.
- [706] arXiv:2505.23799 (replaced) [pdf, html, other]
-
Title: Estimating LLM Consistency: A User Baseline vs Surrogate MetricsSubjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
Large language models (LLMs) are prone to hallucinations and sensitive to prompt perturbations, often resulting in inconsistent or unreliable generated text. Different methods have been proposed to mitigate such hallucinations and fragility -- one of them being measuring the consistency (the model's confidence in the response, or likelihood of generating a similar response when resampled) of LLM responses. In previous work, measuring consistency often relied on the probability of a response appearing within a pool of resampled responses, or internal states or logits of responses. However, it is not yet clear how well these approaches approximate how humans perceive the consistency of LLM responses. We performed a user study (n=2,976) and found current methods typically do not approximate users' perceptions of LLM consistency very well. We propose a logit-based ensemble method for estimating LLM consistency, and we show that this method matches the performance of the best-performing existing metric in estimating human ratings of LLM consistency. Our results suggest that methods of estimating LLM consistency without human evaluation are sufficiently imperfect that we suggest evaluation with human input be more broadly used.
- [707] arXiv:2505.24238 (replaced) [pdf, html, other]
-
Title: MIRAGE: Assessing Hallucination in Multimodal Reasoning Chains of MLLMSubjects: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
Multimodal hallucination in multimodal large language models (MLLMs) restricts the correctness of MLLMs. However, multimodal hallucinations are multi-sourced and arise from diverse causes. Existing benchmarks fail to adequately distinguish between perception-induced hallucinations and reasoning-induced hallucinations. This failure constitutes a significant issue and hinders the diagnosis of multimodal reasoning failures within MLLMs. To address this, we propose the {\dataset} benchmark, which isolates reasoning hallucinations by constructing questions where input images are correctly perceived by MLLMs yet reasoning errors persist. {\dataset} introduces multi-granular evaluation metrics: accuracy, factuality, and LLMs hallucination score for hallucination quantification. Our analysis reveals that (1) the model scale, data scale, and training stages significantly affect the degree of logical, fabrication, and factual hallucinations; (2) current MLLMs show no effective improvement on spatial hallucinations caused by misinterpreted spatial relationships, indicating their limited visual reasoning capabilities; and (3) question types correlate with distinct hallucination patterns, highlighting targeted challenges and potential mitigation strategies. To address these challenges, we propose {\method}, a method that combines curriculum reinforcement fine-tuning to encourage models to generate logic-consistent reasoning chains by stepwise reducing learning difficulty, and collaborative hint inference to reduce reasoning complexity. {\method} establishes a baseline on {\dataset}, and reduces the logical hallucinations in original base models.