论文检索

Xiaodong Zhu, Suting Wang, Yuanming Zheng, Junqi Yang, Yangxu Liao, Yuhong Yang, Weiping Tu, Zhongyuan Wang

Temporal Forgery Localization (TFL) aims to precisely identify manipulated segments in video and audio, offering strong interpretability for security and forensics. While recent State Space Models (SSMs) show promise in precise temporal reasoning, their use in TFL is hindered by ambiguous boundaries, sparse forgeries, and limited long-range modeling. We propose DeformTrace, which enhances SSMs with deformable dynamics and relay mechanisms to address these challenges. Specifically, Deformable Self-SSM (DS-SSM) introduces dynamic receptive fields into SSMs for precise temporal localization. To further enhance its capacity for temporal reasoning and mitigate long-range decay, a Relay Token Mechanism is integrated into DS-SSM. Besides, Deformable Cross-SSM (DC-SSM) partitions the global state space into query-specific subspaces, reducing non-forgery information accumulation and boosting sensitivity to sparse forgeries. These components are integrated into a hybrid architecture that combines the global modeling of Transformers with the efficiency of SSMs. Extensive experiments show that DeformTrace achieves state-of-the-art performance with fewer parameters, faster inference, and stronger robustness.

Carlos Antonio Pinzón, Ehab ElSalamouny, Lucas Massot, Alexis Miller, Héber Hwang Arcolezi, Catuscia Palamidessi

Randomized Response (RR) is a protocol designed to collect and analyze categorical data with local differential privacy guarantees. It has been used as a building block of mechanisms deployed by Big tech companies to collect app or web users' data. Each user reports an automatic random alteration of their true value to the analytics server, which then estimates the histogram of the true unseen values of all users using a debiasing rule to compensate for the added randomness. A known issue is that the standard debiasing rule can yield a vector with negative values (which can not be interpreted as a histogram), and there is no consensus on the best fix. An elegant but slow solution is the Iterative Bayesian Update algorithm (IBU), which converges to the Maximum Likelihood Estimate (MLE) as the number of iterations goes to infinity. This paper bypasses IBU by providing a simple formula for the exact MLE of RR and compares it with other estimation methods experimentally to help practitioners decide which one to use.

Xuancun Lu, Jiaxiang Chen, Shilin Xiao, Zizhi Jin, Zhangrui Chen, Hanwen Yu, Bohan Qian, Ruochen Zhou, Xiaoyu Ji, Wenyuan Xu

Vision-Language-Action (VLA) models revolutionize robotic systems by enabling end-to-end perception-to-action pipelines that integrate multiple sensory modalities, such as visual signals processed by cameras and auditory signals captured by microphones. This multi-modality integration allows VLA models to interpret complex, real-world environments using diverse sensor data streams. Given the fact that VLA-based systems heavily rely on the sensory input, the security of VLA models against physical-world sensor attacks remains critically underexplored. To address this gap, we present the first systematic study of physical sensor attacks against VLAs, quantifying the influence of sensor attacks and investigating the defenses for VLA models. We introduce a novel ``Real-Sim-Real" framework that automatically simulates physics-based sensor attack vectors, including six attacks targeting cameras and two targeting microphones, and validates them on real robotic systems. Through large-scale evaluations across various VLA architectures and tasks under varying attack parameters, we demonstrate significant vulnerabilities, with susceptibility patterns that reveal critical dependencies on task types and model designs. We further develop an adversarial-training-based defense that enhances VLA robustness against out-of-distribution physical perturbations caused by sensor attacks while preserving model performance. Our findings expose an urgent need for standardized robustness benchmarks and mitigation strategies to secure VLA deployments in safety-critical environments.

Pei Huang, Dennis Wei, Omri Isac, Haoze Wu, Min Wu, Clark Barrett

Transformers based on the self-attention mechanism have become foundational models across a wide range of domains, thereby creating an urgent need for effective formal verification techniques to better understand their behavior and ensure safety guarantees. In this paper, we propose two parameterized linear abstract domains for the inner products in the self-attention module, aiming to improve verification precision. The first one constructs symbolic quadratic upper and lower bounds for the product of two scalars, and then derives parameterized affine bounds using tangents. The other one constructs parameterized bounds by interpolating affine bounds proposed in prior work. We evaluate these two parameterization methods and demonstrate that both of them outperform the state-of-the-art approach which is regarded as optimal with respect to a certain mean gap. Experimental results show that, in the context of robustness verification, our approach is able to verify many instances that cannot be verified by existing methods. In the interval analysis, our method achieves tighter results compared to the SOTA, with the strength becoming more pronounced as the network depth increases.

Benjamin Cookson, Soroush Ebadian, Nisarg Shah

Large language models (LLMs) are increasingly used for decision-making tasks where fairness is an essential desideratum. But what does fairness even mean to an LLM? To investigate this, we conduct a comprehensive evaluation of how LLMs perceive fairness in the context of resource allocation, using both synthetic and real-world data. We find that several state-of-the-art LLMs, when instructed to be fair, tend to prioritize improving collective welfare rather than distributing benefits equally. Their perception of fairness is somewhat sensitive to how user preferences are represented, but less so to the real-world context of the decision-making task. Finally, we show that the best strategy for aligning an LLM's perception of fairness to a specific criterion is to provide it as a mathematical objective, without referencing "fairness", as this prevents the LLM from mixing the criterion with its own prior notions of fairness. Our results provide practical insights for understanding and shaping how LLMs interpret fairness in resource allocation problems.

Omar Claflin

Current sparse autoencoder (SAE) approaches to neural network interpretability assume that activations can be decomposed through linear superposition into sparse, interpret-able features. Despite high reconstruction fidelity, SAEs consistently fail to eliminate polysemanticity and exhibit pathological behavioral errors. We propose that neural networks encode information in two complementary spaces com-pressed into the same substrate: feature identity and feature integration. To test this dual encoding hypothesis, we develop sequential and joint-training architectures to capture identity and integration patterns simultaneously. Joint training achieves 41.3% reconstruction improvement and 51.6% reduction in KL divergence errors. This architecture spontaneously develops bimodal feature organization: low squared norm features contributing to integration pathways and the rest contributing directly to the residual. Small nonlinear components (3% of parameters) achieve 16.5% standalone improvements, demonstrating parameter-efficient capture of computational relationships crucial for behavior. Additionally, intervention experiments using 2×2 factorial stimulus designs demonstrated that integration features exhibit selective sensitivity to experimental manipulations and produce systematic behavioral effects on model outputs, including significant interaction effects across semantic dimensions. This work provides systematic evidence for (1) dual encoding in neural representations, (2) meaningful nonlinearly encoded feature interactions, and (3) introduces an architectural paradigm shift from post-hoc feature analysis to integrated computational design, establishing foundations for next-generation SAEs.

Wei Cai, Shujuan Liu, Jian Zhao, Ziyan Shi, Yusheng Zhao, Yuchen Yuan, Tianle Zhang, Chi Zhang, Xuelong Li

Multimodal Large Language Models (MLLMs) are susceptible to the implicit reasoning risk, wherein innocuous unimodal inputs synergistically assemble into risky multimodal data that produce harmful outputs. We attribute this vulnerability to the difficulty of MLLMs maintaining safety alignment through long-chain reasoning.To address this issue, we introduce Safe-Semantics-but-Unsafe-Interpretation (SSUI), the first dataset featuring interpretable reasoning paths tailored for such a cross-modal challenge.A novel training framework, Safety-aware Reasoning Path Optimization (SRPO), is also designed based on the SSUI dataset to align the MLLM's internal reasoning process with human safety values. Experimental results show that our SRPO-trained models achieve state-of-the-art results on key safety benchmarks, including the proposed Reasoning Path Benchmark (RSBench), significantly outperforming both open-source and top-tier commercial MLLMs.

Yang Zhou, Zhenting Sheng, Mingrui Tan, Yuting Song, Jun Zhou, Yu Heng Kwan, Lian Leng Low, Yang Bai, Yong Liu

Effective clinical history taking is a foundational yet underexplored component of clinical reasoning. While large language models (LLMs) have shown promise on static benchmarks, they often fall short in dynamic, multi-turn diagnostic settings that require iterative questioning and hypothesis refinement. To address this gap, we propose Note2Chat, a note-driven framework that trains LLMs to conduct structured history taking and diagnosis by learning from widely available medical notes. Instead of relying on scarce and sensitive dialogue data, we convert real-world medical notes into high-quality doctor-patient dialogues using a decision tree-guided generation and refinement pipeline. We then propose a three-stage fine-tuning strategy combining supervised learning, simulated data augmentation, and preference learning. Furthermore, we propose a novel single-turn reasoning paradigm that reframes history taking as a sequence of single-turn reasoning problems. This design enhances interpretability and enables local supervision, dynamic adaptation, and greater sample efficiency. Experimental results show that our method substantially improves clinical reasoning, achieving gains of +16.9 F1 and +21.0 Top-1 diagnostic accuracy over GPT-4o.

Zijing Zhang, Boning Zhang

Large Language Model (LLM) agents have demonstrated strong potential in complex, interactive decision-making tasks. However, when training LLM agents end-to-end with reinforcement learning (RL), efficiently optimizing agent policies in dynamic environments remains a significant challenge. Existing RL-based LLM agent paradigms commonly organize interactions in a cycle where reasoning is followed by action. In our work, we observe a phenomenon we call Exploration Contraction, where the explicit introduction of a reasoning stage reduces the diversity of actions—quantified by lower action entropy—which in turn limits exploration and leads to premature policy convergence. To address this limitation, we propose Act-before-Reasoning (ActRe), a two-stage RL training framework. In the first stage, we reverse the typical rollout order, prompting the agent to generate actions prior to reasoning, which encourages exploration driven by model intuition. In the second stage, we restore the standard reasoning-then-action order for training and evaluation, ensuring robust and interpretable decision-making. Experiments on the ALFWorld and WebShop benchmarks show that ActRe effectively mitigates exploration contraction, yielding consistently higher task success rates and improved training robustness compared to strong RL baselines. Our analysis underscores the importance of action entropy in the exploration-exploitation trade-off during LLM agent training and provides a practical approach to maintain the benefits of explicit reasoning while promoting sufficient exploration.

Zhenliang Zhang, Xinyu Hu, Xiaojun Wan

Large language models sometimes inadvertently reproduce passages that are copyrighted, exposing downstream applications to legal risk. Most existing studies for inference-time defences focus on surface-level token matching and rely on external blocklists or filters, which add deployment complexity and may overlook semantically paraphrased leakage. In this work, we reframe copyright infringement mitigation as intrinsic semantic-space control and introduce SCOPE, an inference-time method that requires no parameter updates or auxiliary filters. Specifically, the sparse autoencoder (SAE) projects hidden states into a high-dimensional, near-monosemantic space; benefiting from this representation, we identify a copyright-sensitive subspace and clamp its activations during decoding. Experiments on widely recognized benchmarks show that SCOPE mitigates copyright infringement without degrading general utility. Further interpretability analyses confirm that the isolated subspace captures high-level semantics.

Shuyi Zhang, Wei Shi, Sihang Li, Jiayi Liao, Tao Liang, Hengxing Cai, Xiang Wang

Large language models (LLMs) have been widely deployed across numerous fields. Reinforcement Learning from Human Feedback (RLHF) leverages reward models (RMs) as proxies for human preferences to align LLM behaviors with human values, making the accuracy, reliability, and interpretability of RMs critical for effective alignment. However, traditional RMs lack interpretability, offer limited insight into the reasoning behind reward assignments, and are inflexible toward user preference shifts. While recent multidimensional RMs aim for improved interpretability, they often fail to provide feature-level attribution and require costly annotations. To overcome these limitations, we introduce the Sparse Autoencoder-Enhanced Reward Model (SARM), a novel architecture that integrates a pretrained Sparse Autoencoder (SAE) into a reward model. SARM maps the hidden activations of LLM-based RM into an interpretable, sparse, and monosemantic feature space, from which a scalar head aggregates feature activations to produce transparent and conceptually meaningful reward scores. Empirical evaluations demonstrate that SARM facilitates direct feature-level attribution of reward assignments, allows dynamic adjustment to preference shifts, and achieves superior alignment performance compared to conventional reward models.

Mengfei Zhang, Zhenglin Wang

While Large Reasoning Models (LRMs) exhibit remarkable capabilities in complex tasks, they often suffer from excessive redundancy in their chain-of-thought reasoning. This significantly reduces inference efficiency and increases computational costs. We identify that LRM redundancy is not uniformly homogeneous but can be taxonomized according to whether it is destructive to the final answer: destructive redundancy (e.g., logical drift, hallucination amplification) versus non-destructive redundancy (e.g., repetition, over-elaboration). Moreover, LRM's redundant and concise responses exhibit a significant distinction in their hidden layer representation spaces. Based on these insights, we propose CATS (Category-Aware Token-level Steering), a training-free and lightweight method to reduce the redundancy phenomenon. CATS decomposes redundancy into six semantically interpretable characteristic dimensions. By flexibly weighting and combining the differential vectors corresponding to these dimensions, CATS synthesizes a composite intervention vector, enabling zero-parameter intervention in the hidden layers. Experiments across three LRM models and five mathematical reasoning datasets demonstrate that CATS reduces reasoning length by an average of 25% while maintaining or even slightly improving task accuracy. CATS offers a pluggable, training-free, and lightweight solution, making it particularly beneficial for users in low-resource environments.

Jiayuan Zhang, Kaiquan Chen, Zhihao Lu, Enshen Zhou, Qian Yu, Jing Zhang

Web automation uses intelligent agents to perform high-level tasks by mimicking human interactions with webpages. Despite recent advances in LLM-based web agents, efficiently navigating complex, real-world webpages remains challenging due to massive DOM structures (10,000 ~ 100,000 tokens). Current approaches either truncate DOMs—losing vital information—or use inefficient heuristics and separate ranking models, failing to balance precision and scalability. We introduce Prune4Web, a novel paradigm that transforms DOM processing from LLM-based filtering to programmatic pruning. Our key innovation is DOM Tree Pruning Programming, where an LLM generates executable Python scoring programs to dynamically filter DOM elements based on semantic clues from decomposed sub-tasks. This approach eliminates the need for LLMs to process full DOMs, instead delegating traversal and scoring to lightweight, interpretable programs. The result is a 25 ~ 50 times reduction in candidate elements for grounding, enabling precise action localization without attention dilution. Additionally, we propose a data annotation method and a two-turn dialogue training strategy that jointly optimizes Planner, Programmatic Filter, and Grounder in a unified framework. Experiments demonstrate state-of-the-art performance. On our low-level task grounding task, our approach dramatically increases grounding accuracy from 46.80% to 88.28%, highlighting its effectiveness.

Bowen Zhang, Jun Ma, Fuqiang Niu, Li Dong, Jinzhou Cao, Genan Dai

Zero-shot stance detection (ZSSD) seeks to determine the stance of text toward previously unseen targets, a task critical for analyzing dynamic and polarized online discourse with limited labeled data. While large language models (LLMs) offer zero-shot capabilities, prompting-based approaches often fall short in handling complex reasoning and lack robust generalization to novel targets. Meanwhile, LLM-enhanced methods still require substantial labeled data and struggle to move beyond instance-level patterns, limiting their interpretability and adaptability. Inspired by cognitive science, we propose the Cognitive Inductive Reasoning Framework (CIRF), a schema-driven method that bridges linguistic inputs and abstract reasoning via automatic induction and application of cognitive reasoning schemas. CIRF abstracts first-order logic patterns from raw text into multi-relational schema graphs in an unsupervised manner, and leverages a schema-enhanced graph kernel model to align input structures with schema templates for robust, interpretable zero-shot inference. Extensive experiments on SemEval-2016, VAST, and COVID-19-Stance benchmarks demonstrate that CIRF not only establishes new state-of-the-art results, but also achieves comparable performance with just 30% of the labeled data, demonstrating its strong generalization and efficiency in low-resource settings.

Wenzheng Zeng, Mingyu Ouyang, Langyuan Cui, Hwee Tou Ng

Automatic presentation slide generation can greatly streamline content creation. However, since preferences of each user may vary, existing under-specified formulations often lead to suboptimal results that fail to align with individual user needs. We introduce a novel task that conditions paper-to-slides generation on user-specified preferences. We propose a human behavior-inspired agentic framework, SlideTailor, that progressively generates editable slides in a user-aligned manner. Instead of requiring users to write their preferences in detailed textual form, our system only asks for a paper-slides example pair and a visual template—natural and easy-to-provide artifacts that implicitly encode rich user preferences across content and visual style. Despite the implicit and unlabeled nature of these inputs, our framework effectively distills and generalizes the preferences to guide customized slide generation. We also introduce a novel chain-of-speech mechanism to align slide content with planned oral narration. Such a design significantly enhances the quality of generated slides and enables downstream applications like video presentations. To support this new task, we construct a benchmark dataset that captures diverse user preferences, with carefully designed interpretable metrics for robust evaluation. Extensive experiments demonstrate the effectiveness of our framework.

Daojian Zeng, Tianle Li, Jiahao Yang, Jiacai Yi, Xieping Gao, Lincheng Jiang, Tengfei Ma, Xiangxiang Zeng

Multi-objective molecular optimization is a fundamental yet inherently challenging task in drug discovery, as it requires simultaneously optimizing multiple, often conflicting, molecular properties. Although recent deep learning methods have shown promise, they often lack objective-specific specialization and dynamic coordination, making them ineffective in handling competing objectives and difficult to scale in complex, high-dimensional molecular design tasks. Inspired by the division of labor among domain experts in medicinal chemistry, we propose MAMO, a multi-agent framework for molecular design that simulates expert collaboration. Each agent specializes in optimizing a single objective, and their interactions are orchestrated by a central scheduling module that dynamically reallocates tasks based on evaluation feedback. This coordination mechanism enables interpretable and goal-conditioned optimization while adaptively balancing conflicting objectives. Extensive experiments on benchmark datasets demonstrate that MAMO consistently achieves superior performance in both objective quality and Pareto diversity, particularly in scenarios with strong inter-objective conflict. Our results highlight the potential of multi-agent coordination strategies for scalable and conflict-aware molecular design.

Yanke Yu, Jin Li, Ying Sun, Ping Li, Zhefeng Wang, Yi Zheng

Understanding the internal functional organization of Large Language Models (LLMs) is crucial for improving their trustworthiness and performance. However, how LLMs organize different functions into modules remains highly unexplored. To bridge this gap, we formulate a function module discovery problem and propose an Unsupervised LLM Cross-layer MOdule Discovery (ULCMOD) framework that simultaneously disentangles the large set of neurons in the entire LLM into modules while discovering the topics of input samples related to these modules. Our framework introduces a novel objective function and an efficient Iterative Decoupling (IterD) algorithm. Extensive experiments show that our method discovers high-quality, disentangled modules that capture more meaningful semantic information and achieve superior performance in various downstream tasks. Moreover, our qualitative analysis reveals that the discovered modules show function comprehensiveness, function hierarchy, and clear function spatial arrangement within LLMs. Our work provides a novel tool for interpreting LLMs' function modules, filling a critical gap in LLMs' interpretability research.

Shuyang Yu, Jianan Liang, Hui Hu

Optimizing patent claims is a critical yet challenging task, demanding careful balance between maximizing novelty and preserving legal scope. Manual claim drafting is labor-intensive, costly, and inherently inconsistent, while conventional Large Language Models (LLMs) often lack the structured, iterative reasoning essential for precise claim refinement. To address these challenges, we introduce Tree of Claims (ToC), an innovative framework that redefines claim editing as a guided search problem. ToC synergistically integrates Monte Carlo Tree Search (MCTS) with a collaborative multi-agent system, comprising an LLM-based EditorAgent that proposes contextually grounded edits, and an ExaminerAgent that mimics patent examiner critiques through structured, chain-of-thought analyses of novelty and prior art disclosure. Driven by a carefully designed multi-objective reward function, ToC jointly optimizes novelty, scope retention, and semantic coherence. Experimental evaluation on a benchmark of 1145 claims demonstrates that ToC significantly outperforms standard LLMs in zero-shot and few-shot scenarios, achieving an average composite score improvement of 8%, and up to 9% in certain cases. Extensive experiments, including detailed ablation studies, validate ToC’s efficacy in generating superior, legally robust claim revisions. Overall, ToC establishes a transparent, controllable, and interpretable methodology that effectively bridges advanced LLM reasoning capabilities with strategic MCTS planning for structured patent claim optimization.

Rui Yao, Qi Chai, Jinhai Yao, Siyuan Li, Junhao Chen, Qi Zhang, Hao Wang

"Fedspeak", the stylized and often nuanced language used by the U.S. Federal Reserve, encodes implicit policy signals and strategic stances. The Federal Open Market Committee strategically employs Fedspeak as a communication tool to shape market expectations and influence both domestic and global economic conditions. As such, automatically parsing and interpreting Fedspeak presents a high-impact challenge, with significant implications for financial forecasting, algorithmic trading, and data-driven policy analysis. Technically, to enrich the semantic and contextual representation of Fedspeak texts, we incorporate domain-specific reasoning grounded in the monetary policy transmission mechanism. We further introduce a dynamic uncertainty decoding module to assess the confidence of model predictions, thereby enhancing both classification accuracy and model reliability. Experimental results demonstrate that our framework achieves state-of-the-art performance on the policy stance analysis task. Moreover, statistical analysis reveals a significant positive correlation between perceptual uncertainty and model error rates, validating the effectiveness of perceptual uncertainty as a diagnostic signal.

Mutian Yang, Jiandong Gao, Ji Wu

While large language models (LLMs) leverage both knowledge and reasoning during inference, the capacity to distinguish between them plays a pivotal role in model analysis, interpretability, and development. Inspired by dual-system cognitive theory, we propose a cognition attribution framework to decouple the contribution of knowledge and reasoning. In particular, the cognition of LLMs is decomposed into two distinct yet complementary phases: knowledge retrieval (Phase 1) and reasoning adjustment (Phase 2). To separate these phases, LLMs are prompted to generate answers under two different cognitive modes, fast thinking and slow thinking, respectively. The performance under different cognitive modes is analyzed to quantify the contribution of knowledge and reasoning. This architecture is employed to 15 LLMs across 3 datasets. Results reveal: (1) reasoning adjustment is domain-specific, benefiting reasoning-intensive domains (e.g., mathematics, physics, and chemistry) and potentially imparing knowledge-intensive domains. (2) Parameter scaling improves both knowledge and reasoning, with knowledge improvements being more pronounced. Additionally, parameter scaling make LLMs reasoning significantly more prudent, while moderately more intelligent. (3) Knowledge primarily resides in lower network layers, while reasoning operates in higher layers. Our framework not only helps understand LLMs from a "decoupling" perspective, but also provides new insights into existing research, including scaling laws, hierarchical knowledge editing, and limitations of small-scale-LLM reasoning.