论文检索

Deep Learning · Large Language Models

Yuchun Miao, Sen Zhang, Liang Ding, Yuqi Zhang, Lefei Zhang, Dacheng Tao

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.

General Machine Learning

Yuhang Li, Tong Liu, Yangguang Cui, Ming Hu, Xiaoqiang Li

Federated learning (FL) presents a promising strategy for distributed and privacy-preserving learning, yet struggles with performance issues in the presence of heterogeneous data distributions. Recently, a series of works based on sharpness-aware minimization (SAM) have emerged to improve local learning generality, proving to be effective in mitigating data heterogeneity effects.However, most SAM-based methods do not directly consider the global objective and require two backward pass per iteration, resulting in diminished effectiveness.To overcome these two bottlenecks, we leverage the global model trajectory to directly measure sharpness for the global objective, requiring only a single backward pass.We further propose a novel and general algorithm FedGMT to overcome data heterogeneity and the pitfalls of previous SAM-based methods.We analyze the convergence of FedGMT and conduct extensive experiments on visual and text datasets in a variety of scenarios, demonstrating that FedGMT achieves competitive accuracy with state-of-the-art FL methods while minimizing computation and communication overhead. Code is available at https://github.com/harrylee999/FL-SAM.

Applications · Health / Medicine

Yash Shah, Camila Gonzalez, MohammadHassan Abbasi, Qingyu Zhao, Kilian M Pohl, Ehsan Adeli

Confounders are extraneous variables that affect both the input and the target, resulting in spurious correlations and biased predictions. There are recent advances in dealing with or removing confounders in traditional models, such as metadata normalization (MDN), where the distribution of the learned features is adjusted based on the study confounders. However, in the context of continual learning, where a model learns continuously from new data over time without forgetting, learning feature representations that are invariant to confounders remains a significant challenge. To remove their influence from intermediate feature representations, we introduce the Recursive MDN (R-MDN) layer, which can be integrated into any deep learning architecture, including vision transformers, and at any model stage. R-MDN performs statistical regression via the recursive least squares algorithm to maintain and continually update an internal model state with respect to changing distributions of data and confounding variables. Our experiments demonstrate that R-MDN promotes equitable predictions across population groups, both within static learning and across different stages of continual learning, by reducing catastrophic forgetting caused by confounder effects changing over time.

Deep Learning · Other Representation Learning

Adeel Pervez, Efstratios Gavves, Francesco Locatello

We present Mechanistic PDE Networks -- a model for discovery of governing *partial differential equations* from data.Mechanistic PDE Networks represent spatiotemporal data as space-time dependent *linear* partial differential equations in neural network hidden representations. The represented PDEs are then solved and decoded for specific tasks. The learned PDE representations naturally express the spatiotemporal dynamics in data in neural network hidden space, enabling increased modeling power. Solving the PDE representations in a compute and memory-efficient way, however, is a significant challenge. We develop a native, GPU-capable, parallel, sparse and differentiable multigrid solver specialized for linear partial differential equations that acts as a module in Mechanistic PDE Networks. Leveraging the PDE solver we propose a discovery architecture that can discovers nonlinear PDEs in complex settings, while being robust to noise. We validate PDE discovery on a number of PDEs including reaction-diffusion and Navier-Stokes equations.

Reinforcement Learning

Shiqing Gao, Yihang Zhou, Shuai Shao, Haoyu Luo, Yiheng Bing, Jiaxin Ding, Luoyi Fu, Xinbing Wang

Ensuring safety is a critical challenge in applying Reinforcement Learning (RL) to real-world scenarios. Constrained Reinforcement Learning (CRL) addresses this by maximizing returns under predefined constraints, typically formulated as the expected cumulative cost. However, expectation-based constraints overlook rare but high-impact extreme value events in the tail distribution, such as black swan incidents, which can lead to severe constraint violations. To address this issue, we propose the Extreme Value policy Optimization (EVO) algorithm, leveraging Extreme Value Theory (EVT) to model and exploit extreme reward and cost samples, reducing constraint violations. EVO introduces an extreme quantile optimization objective to explicitly capture extreme samples in the cost tail distribution. Additionally, we propose an extreme prioritization mechanism during replay, amplifying the learning signal from rare but high-impact extreme samples. Theoretically, we establish upper bounds on expected constraint violations during policy updates, guaranteeing strict constraint satisfaction at a zero-violation quantile level. Further, we demonstrate that EVO achieves a lower probability of constraint violations than expectation-based methods and exhibits lower variance than quantile regression methods. Extensive experiments show that EVO significantly reduces constraint violations during training while maintaining competitive policy performance compared to baselines.

Deep Learning · Large Language Models

Jiali Cheng, Hadi Amiri

Tool-augmented large language models (LLMs) may need to forget learned tools due to security concerns, privacy restrictions, or deprecated tools. However, ``tool unlearning'' has not been investigated in machine unlearning literature. We introduce this novel task, which requires addressing distinct challenges compared to traditional unlearning: knowledge removal rather than forgetting individual samples, the high cost of optimizing LLMs, and the need for principled evaluation metrics. To bridge these gaps, we propose ToolDelete , the first approach for unlearning tools from tool-augmented LLMs which implements three properties for effective tool unlearning, and a new membership inference attack (MIA) model for evaluation. Experiments on three tool learning datasets and tool-augmented LLMs show that ToolDelete effectively unlearns both randomly selected and category-specific tools, while preserving the LLM's knowledge on non-deleted tools and maintaining performance on general tasks.

General Machine Learning · Evaluation

Evan Frick, Connor Chen, Joseph Tennyson, Tianle Li, Wei-Lin Chiang, Anastasios Angelopoulos, Ion Stoica

Large language model (LLM) evaluations typically rely on aggregated metrics like accuracy or human preference, averaging across users and prompts. This averaging obscures user- and prompt-specific variations in model performance. To address this, we propose Prompt-to-Leaderboard (P2L), a method that produces leaderboards specific to a prompt or set of prompts. The core idea is to train an LLM taking natural language prompts as input to output a vector of Bradley-Terry coefficients which are then used to predict the human preference vote. The resulting prompt-dependent leaderboards allow for unsupervised task-specific evaluation, optimal routing of queries to models, personalization, and automated evaluation of model strengths and weaknesses. Data from Chatbot Arena suggest that P2L better captures the nuanced landscape of language model performance than the averaged leaderboard. Furthermore, our findings suggest that P2L's ability to produce prompt-specific evaluations follows a power law scaling similar to that observed in LLMs themselves. In January 2025, the router we trained based on this methodology achieved the \#1 spot on the Chatbot Arena leaderboard. Our code is available at this GitHub link: https://github.com/lmarena/p2l.

Deep Learning · Generative Models and Autoencoders

Antoni Kowalczuk, Jan Dubiński, Franziska Boenisch, Adam Dziedzic

Image AutoRegressive generation has emerged as a new powerful paradigm with image autoregressive models (IARs) matching state-of-the-art diffusion models (DMs) in image quality (FID: 1.48 vs. 1.58) while allowing for a higher generation speed.However, the privacy risks associated with IARs remain unexplored, raising concerns regarding their responsible deployment. To address this gap, we conduct a comprehensive privacy analysis of IARs, comparing their privacy risks to the ones of DMs as reference points. Concretely, we develop a novel membership inference attack (MIA) that achieves a remarkably high success rate in detecting training images (with a True Positive Rate at False Positive Rate = 1% of 86.38% vs. 6.38% for DMs with comparable attacks). We leverage our novel MIA to provide dataset inference (DI) for IARs, and show that it requires as few as 6 samples to detect dataset membership (compared to 200 for DI in DMs), confirming a higher information leakage in IARs. Finally, we are able to extract hundreds of training data points from an IAR (e.g., 698 from VAR-*d*30). Our results suggest a fundamental privacy-utility trade-off: while IARs excel in image generation quality and speed, they are *empirically* significantly more vulnerable to privacy attacks compared to DMs that achieve similar performance. We release the code at https://github.com/sprintml/privacy_attacks_against_iars for reproducibility.

Applications · Health / Medicine

Feng Wu, Tsai Hor Chan, Fuying Wang, Guosheng Yin, Lequan Yu

Various data modalities are common in real-world applications. (e.g., EHR, medical images and clinical notes in healthcare). Thus, it is essential to develop multimodal learning methods to aggregate information from multiple modalities. The main challenge is appropriately aligning and fusing the representations of different modalities into a joint distribution. Existing methods mainly rely on concatenation or the Kronecker product, oversimplifying interactions structure between modalities and indicating a need to model more complex interactions. Additionally, the joint distribution of latent representations with higher-order interactions is underexplored. Copula is a powerful statistical structure in modelling the interactions between variables, as it bridges the joint distribution and marginal distributions of multiple variables. In this paper, we propose a novel copula modelling-driven multimodal learning framework, which focuses on learning the joint distribution of various modalities to capture the complex interaction among them. The key idea is interpreting the copula model as a tool to align the marginal distributions of the modalities efficiently. By assuming a Gaussian mixture distribution for each modality and a copula model on the joint distribution, our model can also generate accurate representations for missing modalities. Extensive experiments on public MIMIC datasets demonstrate the superior performance of our model over other competitors. The code is anonymously available at https://github.com/HKU-MedAI/CMCM.

Applications · Computer Vision

Zhe Zhang, Mingxiu Cai, Hanxiao Wang, Gaochang Wu, Tianyou Chai, Xiatian Zhu

Unsupervised anomaly detection (UAD) seeks to localize the anomaly mask of an input image with respect to normal samples.Either by reconstructing normal counterparts (reconstruction-based) or by learning an image feature embedding space (embedding-based), existing approaches fundamentally rely on image-level or feature-level matching to derive anomaly scores. Often, such a matching process is inaccurate yet overlooked, leading to sub-optimal detection. To address this issue, we introduce the concept of cost filtering, borrowed from classical matching tasks, such as depth and flow estimation, into the UAD problem. We call this approach CostFilter-AD. Specifically, we first construct a matching cost volume between the input and normal samples, comprising two spatial dimensions and one matching dimension that encodes potential matches. To refine this, we propose a cost volume filtering network, guided by the input observation as an attention query across multiple feature layers, which effectively suppresses matching noise while preserving edge structures and capturing subtle anomalies. Designed as a generic post-processing plug-in, CostFilter-AD can be integrated with either reconstruction-based or embedding-based methods. Extensive experiments on MVTec-AD and VisA benchmarks validate the generic benefits of CostFilter-AD for both single- and multi-class UAD tasks. Code and models will be released at https://github.com/ZHE-SAPI/CostFilter-AD.

Applications · Chemistry, Physics, and Earth Sciences

Yunfei Huang, David S. Greenberg

Neural PDE surrogates can improve the cost-accuracy tradeoff of classical solvers, but often generalize poorly to new initial conditions and accumulate errors over time. Physical and symmetry constraints have shown promise in closing this performance gap, but existing techniques for imposing these inductive biases are incompatible with the staggered grids commonly used in computational fluid dynamics. Here we introduce novel input and output layers that respect physical laws and symmetries on the staggered grids, and for the first time systematically investigate how these constraints, individually and in combination, affect the accuracy of PDE surrogates. We focus on two challenging problems: shallow water equations with closed boundaries and decaying incompressible turbulence. Compared to strong baselines, symmetries and physical constraints consistently improve performance across tasks, architectures, autoregressive prediction steps, accuracy measures, and network sizes. Symmetries are more effective than physical constraints, but surrogates with both performed best, even compared to baselines with data augmentation or pushforward training, while themselves benefiting from the pushforward trick. Doubly-constrained surrogates also generalize better to initial conditions and durations beyond the range of the training data, and more accurately predict real-world ocean currents.

General Machine Learning · Evaluation

Xuehang Guo, Xingyao Wang, Yangyi Chen, Sha Li, Chi Han, Manling Li, Heng Ji

Software engineering (SE) is increasingly collaborative, with developers working together on shared complex codebases. Effective collaboration in shared environments requires participants---whether humans or AI agents---to stay on the same page as their environment evolves. When a collaborator's understanding diverges from the current state---what we term the *out-of-sync* challenge---the collaborator's actions may fail, leading to integration issues. In this work, we introduce **SyncMind**, a framework that systematically defines the *out-of-sync* problem faced by large language model (LLM) agents in collaborative software engineering (CSE). Based on ***SyncMind***, we create **SyncBench**, a benchmark featuring 24,332 instances of agent *out-of-sync* scenarios in real-world CSE derived from 21 popular *GitHub* repositories with executable verification tests. Experiments on ***SyncBench*** uncover critical insights into existing LLM agents' capabilities and limitations. Besides substantial performance gaps among agents (from *Llama-3.1* agents $\leq 3.33\%$ to *Claude-3.5-Sonnet* $\geq 28.18\%$), their consistently low collaboration willingness ($\le 4.86\%$) suggests fundamental limitations of existing LLM in CSE. However, when collaboration occurs, it positively correlates with *out-of-sync* recovery success. Minimal performance differences in agents' resource-aware *out-of-sync* recoveries further reveal their significant lack of resource awareness and adaptability, shedding light on future development of resource-efficient collaborative systems. Our code and data are openly available on our project website: https://xhguo7.github.io/SyncMind/.

Applications

Tingle Li, Baihe Huang, Xiaobin Zhuang, Dongya Jia, Jiawei Chen, Yuping Wang, Zhuo Chen, Gopala Anumanchipalli, Yuxuan Wang

Generating accurate sounds for complex audio-visual scenes is challenging, especially in the presence of multiple objects and sound sources. In this paper, we propose an interactive object-aware audio generation model that grounds sound generation in user-selected visual objects within images. Our method integrates object-centric learning into a conditional latent diffusion model, which learns to associate image regions with their corresponding sounds through multi-modal attention. At test time, our model employs image segmentation to allow users to interactively generate sounds at the object level. We theoretically validate that our attention mechanism functionally approximates test-time segmentation masks, ensuring the generated audio aligns with selected objects. Quantitative and qualitative evaluations show that our model outperforms baselines, achieving better alignment between objects and their associated sounds.

Deep Learning · Attention Mechanisms

Xiang Hu, Zhihao Teng, Jun Zhao, Wei Wu, Kewei Tu

Despite the success of Transformers, handling longer contexts remains challenging due to the limited length generalization and quadratic complexity of self-attention, which often requires post-training with a larger attention window, significantly increasing computational and memory costs. In this paper, we propose a novel attention mechanism based on dynamic context, Grouped Cross Attention (GCA), which can generalize to 1000 $\times$ the pre-training context length while maintaining the ability to access distant information with a constant attention window size. For a given input sequence, we split it into chunks and use each chunk to retrieve top-$k$ relevant past chunks for subsequent text generation. Specifically, unlike most previous works that use an off-the-shelf retriever, our key innovation allows the retriever to learn how to retrieve past chunks that better minimize the auto-regressive loss of subsequent tokens in an end-to-end manner, which adapts better to causal language models.Such a mechanism accommodates retrieved chunks with a fixed-size attention window to achieve long-range information access, significantly reducing computational and memory costs during training and inference. Experiments show that GCA-based models achieve near-perfect accuracy in passkey retrieval for 16M context lengths, which is $1000 \times$ the training length.

Optimization · Discrete and Combinatorial Optimization

Yong Liang Goh, Zhiguang Cao, Yining Ma, Jianan Zhou, Mohammed Haroon Dupty, Wee Sun Lee

Recent advances toward foundation models for routing problems have shown great potential of a unified deep model for various VRP variants. However, they overlook the complex real-world customer distributions. In this work, we advance the Multi-Task VRP (MTVRP) setting to the more realistic yet challenging Multi-Task Multi-Distribution VRP (MTMDVRP) setting, and introduce SHIELD, a novel model that leverages both *sparsity* and *hierarchy* principles. Building on a deeper decoder architecture, we first incorporate the Mixture-of-Depths (MoD) technique to enforce sparsity. This improves both efficiency and generalization by allowing the model to dynamically select nodes to use or skip each decoder layer, providing the needed capacity to adaptively allocate computation for learning the task/distribution specific and shared representations. We also develop a context-based clustering layer that exploits the presence of hierarchical structures in the problems to produce better local representations. These two designs inductively bias the network to identify key features that are common across tasks and distributions, leading to significantly improved generalization on unseen ones. Our empirical results demonstrate the superiority of our approach over existing methods on 9 real-world maps with 16 VRP variants each.

Theory · Probabilistic Methods

Shijie Li, Weiqiang He, Ruobing Bai, Pan Peng

Hierarchical clustering is a widely used method for unsupervised learning with numerous applications. However, in the application of modern algorithms, the datasets studied are usually large and dynamic. If the hierarchical clustering is sensitive to small perturbations of the dataset, the usability of the algorithm will be greatly reduced. In this paper, we focus on the hierarchical $k$ -median clustering problem, which bridges hierarchical and centroid-based clustering while offering theoretical appeal, practical utility, and improved interpretability. We analyze the average sensitivity of algorithms for this problem by measuring the expected change in the output when a random data point is deleted. We propose an efficient algorithm for hierarchical $k$-median clustering and theoretically prove its low average sensitivity and high clustering quality. Additionally, we show that single linkage clustering and a deterministic variant of the CLNSS algorithm exhibit high average sensitivity, making them less stable. Finally, we validate the robustness and effectiveness of our algorithm through experiments.

Reinforcement Learning

Shiqing Gao, Jiaxin Ding, Luoyi Fu, Xinbing Wang

Constrained Reinforcement Learning (CRL) aims to maximize cumulative rewards while satisfying constraints. However, existing CRL algorithms often encounter significant constraint violations during training, limiting their applicability in safety-critical scenarios. In this paper, we identify the underestimation of the cost value function as a key factor contributing to these violations. To address this issue, we propose the Memory-driven Intrinsic Cost Estimation (MICE) method, which introduces intrinsic costs to mitigate underestimation and control bias to promote safer exploration. Inspired by flashbulb memory, where humans vividly recall dangerous experiences to avoid risks, MICE constructs a memory module that stores previously explored unsafe states to identify high-cost regions. The intrinsic cost is formulated as the pseudo-count of the current state visiting these risk regions. Furthermore, we propose an extrinsic-intrinsic cost value function that incorporates intrinsic costs and adopts a bias correction strategy. Using this function, we formulate an optimization objective within the trust region, along with corresponding optimization methods. Theoretically, we provide convergence guarantees for the proposed cost value function and establish the worst-case constraint violation for the MICE update. Extensive experiments demonstrate that MICE significantly reduces constraint violations while preserving policy performance comparable to baselines.

General Machine Learning · Transfer, Multitask and Meta-learning

Kevin Galim, Wonjun Kang, Yuchen Zeng, HYUNG IL KOO, Kangwook Lee

Deep State Space Models (SSMs), such as Mamba (Gu & Dao, 2024), have become powerful tools for language modeling, offering high performance and linear scalability with sequence length. However, the application of parameter-efficient fine-tuning (PEFT) methods to SSM-based models remains largely underexplored. We start by investigating two fundamental questions on existing PEFT methods: (i) How do they perform on SSM-based models? (ii) Which parameters should they target for optimal results? Our analysis shows that LoRA and its variants consistently outperform all other PEFT methods. While LoRA is effective for linear projection matrices, it fails on SSM modules—yet still outperforms other methods applicable to SSMs, indicating their limitations. This underscores the need for a specialized SSM tuning approach. To address this, we propose Sparse Dimension Tuning (SDT), a PEFT method tailored for SSM modules. Combining SDT for SSMs with LoRA for linear projection matrices, we achieve state-of-the-art performance across extensive experiments.

Social Aspects · Privacy

Ce Zhang, Yixin Han, Yafei Wang, Xiaodong Yan, Linglong Kong, Ting Li, Bei Jiang

Randomized response (RR) mechanisms constitute a fundamental and effective technique for ensuring label differential privacy (LabelDP). However, existing RR methods primarily focus on the response labels while overlooking the influence of covariates and often do not fully address optimality. To address these challenges, this paper explores optimal LabelDP procedures using RR mechanisms, focusing on achieving optimal estimation and inference in binary response models. We first analyze the asymptotic behaviors of RR binary response models and then optimize the procedure by maximizing the trace of the Fisher Information Matrix within the $\varepsilon$- and $(\varepsilon,\delta)$-LabelDP constraints. Our theoretical results indicate that the proposed methods achieve optimal LabelDP guarantees while maintaining statistical accuracy in binary response models under mild conditions. Furthermore, we develop private confidence intervals with nominal coverage for statistical inference. Extensive simulation studies and real-world applications confirm that our methods outperform existing approaches in terms of precise estimation, privacy protection, and reliable inference.

Social Aspects · Accountability, Transparency, and Interpretability

Bardh Prenkaj, Efstratios Zaradoukas, Gjergji Kasneci

Counterfactual explainability seeks to uncover model decisions by identifying minimal changes to the input that alter the predicted outcome. This task becomes particularly challenging for graph data due to preserving structural integrity and semantic meaning. Unlike prior approaches that rely on forward perturbation mechanisms, we introduce Graph Inverse Style Transfer (GIST), the first framework to re-imagine graph counterfactual generation as a backtracking process, leveraging spectral style transfer. By aligning the global structure with the original input spectrum and preserving local content faithfulness, GIST produces valid counterfactuals as interpolations between the input style and counterfactual content. Tested on 8 binary and multi-class graph classification benchmarks, GIST achieves a remarkable +7.6% improvement in the validity of produced counterfactuals and significant gains (+45.5%) in faithfully explaining the true class distribution. Additionally, GIST's backtracking mechanism effectively mitigates overshooting the underlying predictor's decision boundary, minimizing the spectral differences between the input and the counterfactuals. These results challenge traditional forward perturbation methods, offering a novel perspective that advances graph explainability.