论文检索

General Machine Learning · Evaluation

Kabir Kang, Steve Mussmann

Standard classification treats all errors equally, but in content moderation, medical screening, and safety-critical applications, mistakes on clear-cut cases are far more costly than errors on ambiguous ones. We propose normalized excess cost (NEC), a metric that weights classification errors by per-example costs and reduces to standard error rate when costs are uniform. Costs can derive from annotator vote margins, distance from decision thresholds, or confidence ratings. Across text, image, and tabular benchmarks, we find that NEC is often substantially lower than error rate—models with 5% error rate can achieve 1.8% NEC—revealing that most mistakes concentrate on ambiguous, low-cost examples. However, incorporating costs into training via loss weighting, sampling strategies, or regression yields inconsistent benefits: improvements appear only when costs are predictable from input features, as in our synthetic control, while real-world datasets show mixed or negligible gains. Our framework provides a practical methodology for deriving and evaluating instance-level misclassification costs, even when cost-sensitive training offers limited benefit.

Deep Learning · Large Language Models

Shihao Wang, Jiahao Chen, Yanqi Pan, Hao Huang, Yichen Hao, Xiangyu Zou, Wen Xia, Chongyang Qiu, Wentao Zhang, PengFei Wang

The prefill stage of long-context Retrieval-Augmented Generation (RAG) is severely bottlenecked by computational overhead. To mitigate this, recent methods assemble pre-calculated KV caches of retrieved RAG documents (by a *user query*) and reprocess selected tokens to recover cross-attention between these pre-calculated KV caches. However, we identify a fundamental ``crowding-out effect'' in current token selection criteria: globally salient but *user-query*-irrelevant tokens saturate the limited recomputation budget, displacing the tokens truly essential for answering the *user query* and degrading inference accuracy. We propose ProphetKV, a user-query-driven KV Cache reuse method for RAG scenarios. ProphetKV dynamically prioritizes tokens based on their semantic relevance to the *user query* and employs a dual-stage recomputation pipeline to fuse layer-wise attention metrics into a high-utility set. By ensuring the recomputation budget is dedicated to bridging the informational gap between retrieved context and the *user query*, ProphetKV achieves high-fidelity attention recovery with minimal overhead. Our extensive evaluation results show that ProphetKV retains 96\%–101\% of full-prefill accuracy with only a 20\% recomputation ratio, while achieving accuracy improvements of 8.8\%–24.9\% on RULER and 18.6\%–50.9\% on LongBench over the state-of-the-art approaches (e.g., CacheBlend, EPIC, and KVShare).

Applications · Computer Vision

Jingjing Liu, Ziye Huang, Zihao Cheng, Zeming Liu, Jiahong Wu, Yuhang Guo, Kehai Chen, Yunhong Wang, Haifeng Wang

While Graphical User Interface (GUI) agents have shown promising performance in automated device interaction, they primarily depend on static parametric knowledge from pre-training or instruction tuning. This reliance fundamentally limits their ability to handle long-tailed tasks that require explicit procedural knowledge absent from model parameters, often forcing agents to resort to inefficient and brittle trial-and-error exploration. To mitigate this limitation, we introduce Proactive Document-Guided Action for GUI agents in dynamic, open-web environments, a novel paradigm that mirrors human problem-solving by enabling agents to autonomously search for relevant documentation to resolve long-tailed tasks. To evaluate agents' capability in this paradigm, we propose DocOS, a benchmark designed to assess document-guided problem solving in fully interactive environments. DocOS requires agents to autonomously navigate a web browser, locate relevant online documentation, comprehend procedural instructions, and faithfully ground them into executable GUI actions. Extensive experiments reveal that progress is strictly constrained by dual bottlenecks: agents struggle to reliably locate relevant information during proactive search and frequently fail to faithfully ground retrieved instructions into precise actions, pointing toward document-guided interaction as a crucial pathway for enabling self-evolving GUI agents in dynamic environments.

Social Aspects · Safety

Jiahui Guang, Yingjie Zhu, Haiyan Wang, Cuiyun Gao, Jing Li, Di Shao, Zhaoquan Gu

Multimodal large language models (MLLMs) inevitably memorize sensitive cross-modal information during pretraining, making post-deployment unlearning crucial for safety. Existing methods often evaluate unlearning based on output deviations, neglecting generation quality, which can lead to hallucinations or rigid refusals, compromising usability and safety. We propose Activation Steering Meets Reinforcement Unlearning for Multimodal Large Language Models (ASRU), a novel, controllable multimodal unlearning framework that considers generation quality as a key evaluation metric. ASRU first constructs an initial model exhibiting refusal-style responses via activation redirection. It then reinforces the refusal boundary using a customized reward function, effectively unlearning target knowledge while retaining non-target knowledge. Experimental results show that ASRU preserves model utility while markedly enhancing unlearning (+24.61%) and generation quality (5.8×), achieving strong generalization and an efficient forget–retain trade-off with minimal retained supervision.

Applications · Health / Medicine

Haoyang Luan, Gufeng Yu, Letian Chen, Zhenran Xiao, Yueshan Huang, Junkun Guo, Yang Yang

The *de novo* generation of high-affinity epitopes tailored to specific major histocompatibility complex (MHC) proteins is a pivotal challenge in computational immunotherapy. However, current methods struggle to effectively integrate the MHC context into the generation process, and often fail to guarantee high binding affinity due to the neglect of discriminative signals from non-binders. To bridge these gaps, we present **EpiCoCo**, a probabilistic framework for **Epi**tope generation via MHC-context **Co**-modeling and **Co**ntrastive affinity learning. EpiCoCo treats the pMHC complex as a dynamic, co-adaptive system by operating on the joint E(3) graph. In addition, we introduce Contrastive Affinity Guidance (CAG), an inference mechanism that leverages the gradient difference between learned high- and low-affinity distributions. CAG actively drives the generation trajectory towards high-affinity manifolds while utilizing repulsive signals to filter out candidates with poor binding potential. Extensive evaluations demonstrate that EpiCoCo achieves a mean binding free energy of -45.20 REU, a 23% improvement over the state-of-the-art, while maintaining high structural plausibility. The results validate that context co-modeling and negative-informed guidance are essential for generating valid, high-potency immunotherapeutics.

Reinforcement Learning · Everything Else

Yannik Schnitzer, Mathias Jackermeier, Alessandro Abate, David Parker

Multi-task reinforcement learning trains generalist policies that can execute multiple tasks. While recent years have seen significant progress, existing approaches rarely provide formal performance guarantees, which are indispensable when deploying policies in safety-critical settings. We present an approach for computing high-confidence guarantees on the performance of a multi-task policy on tasks not seen during training. Concretely, we introduce a new generalisation bound that composes (i) per-task lower confidence bounds from finitely many rollouts with (ii) task-level generalisation from finitely many sampled tasks, yielding a high-confidence guarantee for new tasks drawn from the same arbitrary and unknown distribution. Across state-of-the-art multi-task RL methods, we show that the guarantees are theoretically sound and informative at realistic sample sizes.

Applications · Computer Vision

Ruihao Xia, Yang Tang, Pan Zhou

3D editing—the task of locally modifying the geometry or appearance of a 3D asset—has wide applications in immersive content creation, digital entertainment, and AR/VR. However, unlike 2D editing, it remains challenging due to the need for cross-view consistency, structural fidelity, and fine-grained controllability. Existing approaches are often slow, prone to geometric distortions, or dependent on manual and accurate 3D masks that are error-prone and impractical. To address these challenges, we advance both the data and model fronts. On the data side, we introduce 3DEditVerse, the largest paired 3D editing benchmark to date, comprising 116,309 high-quality training pairs and 1,500 curated test pairs. Built through complementary pipelines of pose-driven geometric edits and foundation model-guided appearance edits, 3DEditVerse ensures edit locality, multi-view consistency, and semantic alignment. On the model side, we propose 3DEditFormer, a 3D-structure-preserving transformer. By enhancing image-to-3D generation with dual-guidance attention and time-adaptive gating, 3DEditFormer disentangles editable regions from preserved structure, enabling precise and consistent edits without requiring auxiliary 3D masks. Extensive experiments demonstrate that our framework outperforms state-of-the-art baselines both quantitatively and qualitatively, establishing a new standard for practical and scalable 3D editing. Dataset and code will be released. Project: https://anonymousresearch37.github.io/3DEditFormer/

Deep Learning · Theory

Léa Bohbot, Cyril Letrouit, Gabriel Peyré, François-Xavier Vialard

As context windows in large language models continue to expand, it is essential to characterize how attention behaves at extreme sequence lengths. We introduce token-sample complexity: the rate at which attention computed on $n$ tokens converges to its infinite-token limit. We estimate finite-$n$ convergence bounds at two levels: pointwise uniform convergence of the attention map, and convergence of moments for the transformed token distribution. For compactly supported (and more generally sub-Gaussian) distributions, our first result shows that the attention map converges uniformly on a ball of radius $R$ at rate $C(R)/\sqrt{n}$, where $C(R)$ grows exponentially with $R$. For large $R$, this estimate loses practical value, and our second result addresses this issue by establishing convergence rates for the moments of the transformed distribution (the token output of the attention layer). In this case, the rate is $C'(R)/n^{\beta}$ with $\beta<\tfrac{1}{2}$, and $C'(R)$ depends polynomially on the size of the support of the distribution. The exponent $\beta$ depends on the attention geometry and the spectral properties of the tokens distribution. We also examine the regime in which the attention parameter tends to infinity and the softmax approaches a hardmax, and in this setting, we establish a logarithmic rate of convergence. Experiments on synthetic Gaussian data and real BERT models on Wikipedia text confirm our predictions.

Applications · Everything Else

Tianpeng Bu, Xin Liu, Qihua Chen, Hao Jiang, Shurui Li, hongtao duan, Lu Jiang, lulu hu, Bin Yang, Minying Zhang

While GUI agents have advanced rapidly, they often lack the robustness to recover from their own errors, hindering real-world deployment. To bridge this gap at both the evaluation and data levels, we introduce GUI-RobustEval and propose Robustness-driven Trajectory Synthesis. GUI-RobustEval containing 1,216 executable test cases that systematically measure error recovery capabilities across a broad and realistic spectrum of error modes. At the data level, RoTS is a scalable synthesis framework that creates 800k high quality data via a tree-based pipeline that proactively discovers diverse error modes and synthesizes corresponding recovery steps. Our two models, RoTS-7B and RoTS-32B, fine-tuned on our dataset, both demonstrate significant gains on GUI-RobustEval and traditional GUI benchmarks. Notably, RoTS-32B achieves state-of-the-art performance on OSWorld, with a 47.4% success rate and a 33.8% All-Pass@4 score, indicating that the enhanced long-horizon error recovery ability synergistically boosts robustness and overall performance. We will release our benchmark, dataset, and models to facilitate future research.

Bowen Xu, Shaoyu Wu, Hao Jiang, Kai Liu, Xin Chen, lulu hu, Bin Yang

Effective tool use and reasoning are essential capabilities for large reasoning models (LRMs) to address complex real-world problems. Through empirical analysis, we identify a prevalent "Lazy Reasoning" phenomenon, where LRMs frequently engage in repetitive and meaningless reflective reasoning. This occurs primarily due to their inadequate ability to decompose tasks when reasoning in complex tool use scenarios. To address this, we propose a two-stage training framework D-CORE ( Decomposing tasks and Composing Reasoning processes) that first incentivize the LRM’s task decomposition reasoning capability via self-distillation, followed by diversity-aware reinforcement learning (RL) to restore LRM's reflective reasoning capability. D-CORE achieves robust tool-use improvements across diverse benchmarks and model scales. Experiments on BFCLv3 demonstrate superiority of our method: D-CORE-8B reaches 77.7% accuracy, surpassing the best-performing 8B model by 5.7%. Meanwhile, D-CORE-14B establishes a new state-of-the-art at 79.3%, outperforming 70B models despite being 5× smaller. The source code and data sample are in the supplementary material.

Deep Learning · Large Language Models

Hao Jiang, Shurui Li, Tianpeng Bu, Bowen Xu, Xin Liu, Qihua Chen, hongtao duan, lulu hu, Bin Yang, Minying Zhang

Recent advances in online reinforcement learning (RL) for large language models (LLMs) have demonstrated promising performance in complex reasoning tasks. However, they often exhibit an imbalanced exploration–exploitation trade-off, resulting in unstable optimization and sub-optimal performance. We introduce IB-Score, a novel metric grounded in Information Bottleneck (IB) theory that evaluates policy’s exploration-exploitation balance by quantifying the trade-off between step-level reasoning diversity and mutual information shared with the correct answer. Analysis based on IB-Score shows that popular online RL approaches (e.g., GRPO) with common regularization methods fail to consistently maintain balance during training with suboptimal results. To address this, we propose Information Bottleneck-driven Tree-based Policy Optimization (IB-TPO), a principled framework that formulates IB-Score as a fine-grained optimization objective and utilizes a novel IB-guided tree sampling strategy that not only improves the efficiency of online sampling with 50\% more trajectories under same token budget, but also reuses the tree structure for effective IB-Score Monte Carlo estimation. Extensive experiments across standard benchmarks show that our method significantly outperforms GRPO baseline by 2.9% to 3.6% and also outperforms other state-of-the-art online RL approaches.

Deep Learning · Large Language Models

Ivo Petrov, Jasper Dekoninck, Martin Vechev

Large language models (LLMs) have recently shown strong performance on mathematical benchmarks. At the same time, they are prone to hallucination and sycophancy, often providing convincing but flawed proofs for incorrect mathematical statements provided by users. This significantly limits the applicability of LLMs in theorem proving, as verification of these flawed proofs must be done manually by expert mathematicians. However, existing benchmarks that measure sycophancy in mathematics are limited: they focus solely on final-answer problems, rely on very simple and often contaminated datasets, and construct benchmark samples using synthetic modifications that create ill-posed questions. To address these issues, we introduce BrokenMath, the first benchmark for evaluating sycophantic behavior in LLMs within the context of natural language theorem proving. BrokenMath is built from advanced 2025 competition problems, which are perturbed with an LLM to produce false statements and subsequently refined through expert review. We evaluate state-of-the-art LLMs and agentic systems and find that sycophancy is widespread, with the best model, GPT-5, producing sycophantic answers 29% of the time. We further investigate several mitigation strategies, including test-time interventions and supervised fine-tuning on curated sycophantic examples. These approaches reduce, but do not eliminate, sycophancy.

Deep Learning · Large Language Models

Songwen Zhao, Danqing Wang, Kexun Zhang, Jiaxuan Luo, Zhuo Li, Lei Li

Vibe coding is a new programming paradigm in which human engineers instruct large language model (LLM) agents to complete complex coding tasks with little supervision. Although it is increasingly adopted, are vibe coding outputs really safe to deploy in production? To answer this question, we propose SUSVIBES, a benchmark consisting of 200 feature-request software engineering tasks from real-world open-source projects, which, when given to human programmers, led to vulnerable implementations. We evaluate multiple widely used coding agents with frontier models on this benchmark. Disturbingly, all agents perform poorly in terms of software security. Although 61% of the solutions from SWE-Agent with Claude 4 Sonnet are functionally correct, only 10.5% are secure. Further experiments demonstrate that preliminary security strategies, such as augmenting the feature request with vulnerability hints, cannot mitigate these security issues. Our findings raise serious concerns about the widespread adoption of vibe-coding, particularly in security-sensitive applications.

Deep Learning · Large Language Models

Jingxuan Xu, Ken Deng, Weihao Li, Songwei Yu, Haoyang Huang, Yifan Yao, Huaixi Tang, Zhiyi Lai, Kepeng Lei, Zizheng Zhan 等

Evaluating large language models (LLMs) for software engineering has been limited by narrow task coverage, language bias, and insufficient alignment with real-world developer workflows. Existing benchmarks often focus on algorithmic problems or Python-centric bug fixing, leaving critical dimensions of software engineering underexplored. To address these gaps, we introduce SWE-Compass, a comprehensive benchmark that unifies heterogeneous code-related evaluations into a structured and production-aligned framework. SWE-Compass spans 8 task types, 8 programming scenarios, and 10 programming languages, with 2000 high-quality instances curated from authentic GitHub pull requests and refined through systematic filtering and validation. We benchmark ten state-of-the-art LLMs under two agentic frameworks, SWE-Agent and Claude Code, revealing a clear hierarchy of difficulty across task types, languages, and scenarios. Moreover, by aligning evaluation with real-world developer practices, we hope SWE-Compass can provide a rigorous and reproducible foundation for diagnosing and advancing agentic coding capabilities in large language models.

Deep Learning · Algorithms

Guangrui Fan, DanDan Liu, AZNUL SABRI, Pan Lihu

Preference-based reward modeling is a core component of RLHF and DPO pipelines. In practice, the humans providing preference feedback are rarely an i.i.d. sample: recruitment and exposure often follow social, institutional, or spatial structure, inducing non-uniform inclusion probabilities that correlate with graph centrality. We formalize preference learning with *network-sampled* annotators and show that identity-agnostic scalar reward modeling implicitly represents an inclusion-weighted welfare, over-representing structurally central communities when the inclusion distribution $q$ differs from a designer-chosen target weighting $\pi$. We propose Graph-Preference Learning, which combines (i) a graph-personalized reward model that shares statistical strength across neighboring annotators and (ii) graph-balanced aggregation that computes stabilized importance weights to target $\pi$. Our analysis characterizes the induced welfare represented by the learned aggregate reward and bounds its deviation from the target in terms of weight mismatch, reward-model approximation, and finite-sample effects. Experiments on synthetic graphs and a *semi-synthetic* case study on the LMArena preference dataset, where biased inclusion is *induced* via graph-based sampling, demonstrate up to 62% reduction in target-welfare recovery error and 17% reduction in cross-language performance gaps under biased inclusion.

Applications · Time Series

Hanyin Cheng, Jingrong Zhou, Yang Shu, Chenjuan Guo

Probabilistic forecasting with exogenous variables is vital for decision-making but remains underexplored compared to deterministic methods. We propose KITE, a knowledge-guided probabilistic modeling framework designed to bridge this gap by addressing two key bottlenecks: (1) topological disparity in sampling initialization and (2) spurious covariate correlations during the iterative conditional generation process. KITE introduces a *History-Conditional Manifold* to construct an informative source distribution from historical dynamics, effectively anchoring the starting point closer to the target space. Additionally, a *Knowledge-Guided Conditioning* module is developed to regularize variable interactions using statistical priors, suppressing spurious correlations and enhancing the robustness of covariate conditioning. Extensive experiments demonstrate that KITE outperforms state-of-the-art methods in both deterministic and probabilistic forecasting.

Deep Learning · Generative Models and Autoencoders

Renzo Soatto, Anders Hoel, Greycen Ren, Shorna Alam, Stephen Bates, Nikolaos Daskalakis, Caroline Uhler, Maria Skoularidou

Diffusion models have excelled at generative tasks for both continuous and token-based domains, but their application to discrete ordinal data remains underdeveloped. We present \emph{CountsDiff}, a diffusion framework designed to natively model distributions on the natural numbers. CountsDiff extends the Blackout diffusion framework by simplifying its formulation through a direct parameterization in terms of a survival probability schedule and an explicit loss weighting. This introduces flexibility through design parameters with direct analogues in existing diffusion modeling frameworks. Beyond this reparameterization, CountsDiff introduces features from modern diffusion models, previously absent in counts-based domains, including continuous-time training, classifier-free guidance, and churn/remasking reverse dynamics that allow non-monotone reverse trajectories. We propose an initial instantiation of CountsDiff and validate it on natural image datasets (CIFAR-10, CelebA), demonstrating the benefits of the proposed design space and that the framework scales to complex, high-dimensional data domains. We then highlight biological count assays as a natural use case, evaluating CountsDiff on single-cell RNA-seq imputation in a fetal cell and heart cell atlas. Remarkably, we find that even this simple instantiation matches or surpasses the performance of a state-of-the-art discrete generative model and leading RNA-seq imputation methods, while leaving substantial headroom for further gains through optimized design choices in future work.

Deep Learning · Generative Models and Autoencoders

Jiankai Zuo, Yu Zhang, Yang Zhang, Zihao Yao, YAYING ZHANG

Learning on graphs with missing node attributes is a prevalent yet challenging problem in real-world scenarios, as graph neural networks (GNNs) typically rely on complete attribute information. Existing solutions often employ adversarial learning in a shared latent space to align graph structure and attributes. However, these methods frequently suffer from training instability and mode collapse, failing to fully capture the complex, multi-modal joint distribution of topology and features. To address these limitations, we present GLAD (Graph Latent Attribute Diffusion with Bidirectional Alignment), a novel generative framework for robust node attribute completion. GLAD leverages the strong generative capabilities of diffusion models to learn the conditional distribution of attributes given the graph structure within a decoupled latent space. Unlike previous unidirectional approaches, GLAD introduces a robust bidirectional alignment mechanism. Specifically, we incorporate a structure reconstruction constraint during training and structure-aware classifier-free guidance during sampling, ensuring that generated attributes are not only plausible but also maintain strict topological consistency with the underlying graph. Theoretically, we show that GLAD maximizes a tighter variational lower bound on the joint log-likelihood compared to GAN-based predecessors, leading to superior mode coverage. Extensive experiments on standard and large-scale benchmarks demonstrate that GLAD significantly outperforms state-of-the-art baselines in both attribute recovery quality and downstream task performance.

Reinforcement Learning · Batch/Offline

Lukas Schäfer, Pallavi Choudhury, Abdelhak Lemkhenter, Chris Lovett, Somjit Nath, Luis França, Matheus Mendonca, Alex Lamb, Riashat Islam, Siddhartha Sen 等

Behavior cloning (BC) is a practical offline imitation learning method, but it often fails when expert demonstrations are limited. Recent works have introduced a class of architectures named predictive inverse dynamics models (PIDM) that combine a future state predictor with an inverse dynamics model (IDM). While PIDM often outperforms BC, the reasons behind its benefits remain unclear. In this paper, we provide a theoretical explanation: PIDM introduces a bias-variance tradeoff. While predicting the future state introduces bias, conditioning the IDM on the prediction can significantly reduce variance. We establish conditions on the state predictor bias for PIDM to achieve lower prediction error and higher sample efficiency than BC, with the gap widening when additional data sources are available. We validate the theoretical insights empirically in 2D navigation tasks, where BC requires up to five times (three times on average) more demonstrations than PIDM to reach comparable performance; and in a complex 3D environment in a modern video game with high-dimensional visual inputs and stochastic transitions, where BC requires over 66\% more samples than PIDM.

Applications · Health / Medicine

Hongyu Shi, Kaizhong Zheng, WS Zhai, Shuai Jiang, Liangjun Chen, Badong Chen

Multi-modal neuroimaging diagnosis must integrate cross-modal agreement with modality-specific complementarity, yet in real multi-site cohorts these signals are frequently entangled with site- and cohort-dependent correlations, yielding shortcut-driven predictions, fragile transfer, and limited interpretability. We propose Structured Multi-modal Graph Disentanglement (SMGD), which explicitly factorizes multi-modal graph representations into four components with distinct roles: shared diagnostic evidence, complementary diagnostic evidence, incidental cross-modal agreement, and modality-specific non-robust correlations. SMGD is realized as geometry-driven structure learning: under a mild distributional assumption, we develop mini-batch estimable surrogate regularizers that shape subspace organization and cross-modal relations, enforcing semantic consistency through relational geometry rather than centroid coincidence while suppressing confounded dependencies. Experiments on large multi-site datasets (ABIDE-I, SRPBS) show improved in-domain diagnosis and more reliable cross-dataset generalization under modality gap, without expert-crafted features. Code is available at: \url{https://anonymous.4open.science/r/anonymous_ICML_2026/README.md}.