论文检索

Deep Learning · Large Language Models

Quanxin Liu, Yijun Mo, Ruida Xu, Jianwei Zhong, Changhu Chen, Rui Hao

End-to-end data science agent workflows involve tightly coupled sub-processes with strong dynamic dependencies, posing a challenging long-horizon orchestration problem. Existing frameworks primarily rely on static, chain-like execution plans, which are prone to error propagation from early stages—often causing reasoning chain collapse and task failure, resulting in fragile inference and poor cost-effectiveness. To address these issues, we propose $\text{R}^3$DAO, a reactive data agent orchestration framework based on feedback-driven topology evolution, aiming to build a dynamic evolutionary closed-loop of "hierarchical exploration, iterative recovery, and empirical convergence." First, we introduce a dynamic hierarchical task network that recursively decomposes global intent into macro-logical anchors and micro-operators, enabling low-cost exploration through dimensionality reduction in the logical space. Second, we establish a reactive topology reconfiguration mechanism that leverages semantic reflection to map execution anomalies into diagnostic signals, replacing costly global resets with localized topological optimization for resilient self-healing. Finally, semantic experience distillation implements a dual-loop accumulation that compresses long-horizon trajectories into structured prior, steering execution efficiency toward the optimal regime. Evaluations on the MLE-bench show that $\text{R}^3$DAO achieves a 77.36\% improvement in success rate over advanced R\&D-Agent while maintaining competitive task scores. Notably, $\text{R}^3$DAO compresses the average execution time by 36$\times$ and limits token consumption to just 104k per task, showcasing superior reliability, efficiency, and cost-effectiveness.

Social Aspects · Safety

Zirui Hu, Zheng Zhang, Yingjie Wang, Dacheng Tao

Large language models (LLMs) are vulnerable to malicious inputs, posing serious risks in high-stakes applications. While existing detection-based methods have shown strong empirical performance, they fail to explicitly control the false acceptance rate (FAR), a critical safety metric in sensitive scenarios. This limitation is further compounded by two key challenges: the absence of access to known malicious samples and the dynamic, online nature of real-world data streams. To address these challenges, we propose \textit{OSCS}, a novel framework designed to enable online FAR control without relying on malicious calibration samples. OSCS leverages detection scores from existing defenses and employs recursive density estimation to infer benign likelihoods directly from the test stream. This approach allows OSCS to make real-time accept/reject decisions while adhering to a user-specified FAR threshold. We provide theoretical guarantees demonstrating that OSCS effectively controls the FAR, with only a vanishing excess term under mild conditions. Extensive experiments on backdoor attacks and jailbreak scenarios further validate OSCS's effectiveness, showing that it consistently achieves robust FAR control across a variety of tasks and attack settings. These results underscore the practicality of OSCS for ensuring safety in critical LLM deployments.

Zexue He, Yu Wang, Churan Zhi, Yuanzhe Hu, Tzu-Ping Chen, Lang Yin, Ze Chen, Tong Wu, Siru Ouyang, Tom Tang 等

Existing evaluations of agents with memory typically assess **memorization** and **action** in isolation. One class of benchmarks evaluates memorization by testing recall of past conversations or text but fails to capture how memory is used to guide future decisions. Another class focuses on agents acting in single-session tasks without the need for long-term memory. However, in realistic settings, memorization and action are tightly coupled: agents acquire memory while interacting with the environment, and subsequently rely on that memory to solve future tasks. To capture this setting, We introduce MEMORYARENA, a unified evaluation gym for benchmarking agent memory in multi-session Memory-Agent-Environment loops. The benchmark consists of human-crafted agentic tasks with explicitly interdependent subtasks, where agents must learn from earlier actions and feedback by distilling experiences into memory, and subsequently use that memory to guide later actions to solve the overall task. MEMORYARENA supports evaluation across web navigation, preference-constrained planning, progressive information search, and sequential formal reasoning, and reveals that agents with near-saturated performance on existing long-context memory benchmarks like LoCoMo perform poorly in our agentic setting, exposing a gap in current evaluations for agents with memory.

General Machine Learning · Causality

Sawal Acharya, Terry Zhang, Andrew Kim, Anahita Haghighat, Xianlin Sun, Pepijn Cobben, Rahul Shrestha, Maximilian Mordig, Jacob Emmerson, Furkan Danisman 等

Identifying and estimating causal relationships from data is an important component of the scientific research process because it enables researchers to understand how variables affect one another. While large language models (LLMs) show potential for assisting research workflows, their ability to perform causal inference in empirical studies remains underexplored, despite the importance of causality in domains such as medicine and public policy. To address this gap, we introduce CauSciBench, a benchmark for evaluating LLMs' ability to perform end-to-end causal inference autonomously to answer causal questions that arise in empirical research. CauSciBench contains over 300 evaluation tasks derived from real-world studies across multiple disciplines, synthetic scenarios, and textbook datasets. Prior causal inference benchmarks primarily evaluate whether LLMs can implement user-specified methods. In contrast, CauSciBench evaluates performance across the full causal analysis pipeline, including variable selection, method selection, causal effect estimation, and statistical interpretation. We evaluate seven frontier models using several test-time scaling strategies, including Chain-of-Thought, Program-of-Thought, and ReAct. Results show a clear performance gap between real-world and synthetic settings, highlighting limitations in current agentic capabilities for data-driven causal analysis.

Deep Learning · Large Language Models

Kefan Song, Yanjun Qi

Autonomous CLI agents can now execute hundreds of actions across multi-hour sessions: writing code, executing shell commands, browsing the web, and managing cloud infrastructure, all with minimal human oversight. Does greater autonomy invite greater risk? We introduce ANCHOR, an automated auditing framework that stress-tests CLI agents on illegal tasks grounded in public US court cases. ANCHOR deploys an adversarial auditor agent fine-tuned on dark personality data using supervised and reinforcement learning. This auditor roleplays persistent malicious users who decompose tasks, reframe requests upon refusal, and adapt strategies across multi-turn interactions. Evaluating frontier models, we find that while they often refuse illegal tasks when prompted directly, compliance rates reaches 100% under persistent malicious interaction. When agents comply, they frequently exceed user requests, autonomously building infrastructure for large-scale harm, including catastrophic risk scenarios such as large-scale financial fraud and bioweapon development. These findings demonstrate that current alignment techniques are insufficient for autonomous agents and underscore the need for safety evaluations against persistent, adaptive malicious users.

Deep Learning · Large Language Models

ZHIXUAN WU, Quanxing Zha, Teng Wang, Genbao Xu, Wenyuan Gu, Wei Rao, Nan Ma, Bo Cheng, Soujanya Poria

Video understanding requires identifying and reasoning over semantically discriminative visual objects across frames, yet existing object-agnostic solutions struggle to effectively handle substantial object variations over time. To address this, we introduce Chain-of-Glimpse, a search-guided progressive object-grounded reasoning framework that explicitly anchors each reasoning step to specific visual evidence regions, enabling compositional and multi-step decision-making. Formally, Chain-of-Glimpse formulates video reasoning as a step-by-step process that incrementally builds spatially grounded traces around task-relevant visual objects, thereby mitigating over-reliance on saliency-driven cues. Specifically, Chain-of-Glimpse features a search-guided controller, optimized via reinforcement learning with a format reward that significantly incentivizes grounding capability, to iteratively ground visual evidence regions and form reliable reasoning trajectories, yielding accurate and interpretable multi-step decisions. Extensive evaluations on both in-domain NExTQA and out-of-domain Video-Holmes, CG-Bench-Reasoning, and VRBench benchmarks demonstrate consistent performance gains, robustness and generalization of Chain-of-Glimpse across diverse video reasoning tasks.

Deep Learning · Large Language Models

Mehryar Mohri, Yutao Zhong

Preference learning has become the foundation of aligning Large Language Models (LLMs) with human intent. Popular methods, such as Direct Preference Optimization (DPO), minimize surrogate losses as proxies for the intractable pairwise ranking loss. However, we demonstrate that for the equicontinuous hypothesis sets typical of neural networks, these standard surrogates are theoretically inconsistent, yielding vacuous generalization guarantees. To resolve this, we formulate LLM alignment within a margin-shifted ranking framework. We derive rigorous $H$-consistency bounds that depend on enforcing a separation margin $\gamma$. Crucially, we extend this to Structure-Aware $H$-consistency, introducing a novel objective (SA-DPO) that adapts the margin based on the semantic distance between responses to handle synonyms and hard pairs. Finally, we analyze the trade-off between consistency and model limitations via the Margin-Capacity Profile, proving that heavy-tailed surrogates (such as the Polynomial Hinge family) offer superior consistency guarantees for capacity-bounded models compared to the standard logistic loss used in DPO.

Philip Jordan, Maryam Kamgarpour

We study the existence and computation of Nash equilibria in concave games where the players' admissible strategies are subject to shared coupling constraints. Under playerwise concavity of constraints, we prove existence of Nash equilibria. Our proof leverages topological fixed point theory and novel structural insights into the contractibility of feasible sets, and relaxes strong assumptions for existence in prior work. Having established existence, we address the question of whether in the presence of coupling constraints, playerwise independent learning dynamics have convergence guarantees. We address this positively for the class of potential games by designing a convergent algorithm. To account for the possibly nonconvex feasible region, we employ a log barrier regularized gradient ascent with adaptive stepsizes. Starting from an initial feasible strategy profile and under exact gradient feedback, the proposed method converges to an $\epsilon$-approximate constrained Nash equilibrium within $\mathcal{O}(\epsilon^{-3})$ iterations.

Optimization · Stochastic

Hojin Ko, Jeonggyu Huh

Most value-based and actor-critic reinforcement learning methods rely on Bellman-style recursions, yet these recursions collapse under non-exponential discounting common in human preferences and survival processes. We show the breakdown is structural: exponential discounting sits at a fragile intersection of multiplicativity and time homogeneity, and violating either property breaks standard dynamic programming. To overcome this, we propose **Pontryagin-Guided Direct Policy Optimization (PG-DPO)**, a variational framework that abandons recursion and couples the Pontryagin Maximum Principle with Monte Carlo rollouts via an *Adjoint-MC projection* enforcing pointwise Hamiltonian maximization. Across multidimensional hyperbolic and survival-discount benchmarks, PG-DPO improves accuracy and stability where equation-driven solvers and critic-based baselines diverge.

Deep Learning · Generative Models and Autoencoders

Yichen Wang, Chenghao Yang, Tenghao Huang, Muhao Chen, Jonathan May, Mina Lee

Alignment has greatly improved large language models (LLMs)’ output quality at the cost of diversity, yielding highly similar outputs across generations, especially in open-ended generation tasks. We propose Base-Aligned Model Collaboration (BACo), an inference-time token-level model collaboration framework that dynamically combines a base LLM with its aligned counterpart to optimize diversity and quality. Using uncertainty and content-based signals, BACO employs routing strategies to determine, at each token, which model to decode from. Prior diversity-promoting methods often improve diversity at the expense of quality or require expensive decoding or post-training. In contrast, BACO achieves both high diversity and quality post hoc within a single pass, while offering strong controllability. We introduce a family of effective routing strategies and evaluate them across three open-ended generation tasks with 13 diversity and quality metrics. BACo consistently surpasses state-of-the-art inference-time baselines. With our best router, BACo achieves a 21.3% joint improvement in diversity and quality, which is further supported by human evaluations. Overall, our results demonstrate that collaboration between base and aligned models provides an effective and controllable mechanism for optimizing the diversity-quality trade-off.

General Machine Learning · Evaluation

Haocheng Wang, Baiyu Huang, Yingjia Wan, Xiao Zhu, Xiaoyang Liu, Yinya Huang, Zhijiang Guo

Autoformalization—translating mathematical problems from natural language into formal proof assistant code—is essential for rigorous machine reasoning. However, existing evaluation frameworks provide only opaque binary verdicts or scalar scores, offering no interpretable insight into where or why translations fail. This opacity severely limits both human understanding and automated system improvement. To bridge this gap, we introduce \textbf{FormalRX}, a comprehensive diagnostic evaluation framework that transforms autoformalization assessment from black-box judgments into actionable feedback. At its core is \textbf{\textsc{Sci} Error Taxonomy}, a hierarchical classification scheme decomposing autoformalization errors into 28 distinct categories with strict priority ordering. Building on this taxonomy, FormalRX provides four critical diagnostic capabilities: (1) alignment verdicts, (2) error categorization, (3) error localization, and (4) correction. Our diagnostic model % specialized diagnostic models and evaluate them across all tasks. Our model FormalRX-8B, achieves F1-scores of 0.88 (verdict) and 0.71 (categorization), along with accuracies of 0.75 (localization) and 0.73 (correction), substantially outperforming both general-purpose LLMs and specialized baselines. By connecting evaluation with actionable insights, FormalRXenables systematic diagnosis and improvement of autoformalization systems.

Applications · Language, Speech and Dialog

Sanidhya Vijayvargiya, Vijay Viswanathan, Graham Neubig

Humans do not always express what they need perfectly. Helpful assistants must be able to ask for clarification to handle real-world user commands which may be underspecified or poorly stated. Yet, optimal clarification remains challenging. The space of possible tasks is large, and not all missing information is equally valuable. We study real software engineering tasks. These tasks require the user to provide many details, some of which are more important than others, and an accurate assistant should be able to determine which clarification questions to ask when necessary information is missing. We systematically quantify which types of information in prompts types most impact task success and what types of clarification questions enable productive answers by simulated users. Using Shapley attribution and distributional comparisons, we identify two learnable properties of effective clarification: task relevance (which information impacts success) and user answerability (what users can realistically provide). We operationalize these as multi-stage reinforcement learning rewards to train an 8B parameter module matching GPT-5's performance while generating 41\% fewer questions. Our work demonstrates that effective clarification emerges from grounding reward design in empirical analysis of information impact and user answerability, establishing a generalizable reward framework applicable across interactive task domains.

Xinrui Wang, Shao-Yuan Li, Bartłomiej Twardowski, Alexandra Gomez-Villa, Songcan Chen

Online Continual Learning (OCL) aims to learn from endless non\text{-}stationary data streams, yet most existing methods assume a flat label space and overlook the hierarchical organization of real\text{-}world concepts that evolves both horizontally (sibling classes) and vertically (coarse or fine categories). To better reflect this context, we introduce a new problem setting, DHOCL (Online Continual Learning from Dynamic Hierarchies), where taxonomies evolve across granularities and each sample provides supervision at a single hierarchical level. In this setting, we find two fundamental issues: (i) partial supervision under mixed granularities provides only point-wise signals over an evolving path-wise hierarchy, which constrains plasticity and undermines cross-level semantic consistency, and (ii) the dynamically evolving hierarchies induce granularity-dependent interference, destabilizing popular replay and regularization mechanisms and thereby exacerbating catastrophic forgetting. To tackle these issues, we propose HALO (Hierarchical Adaptive Learning with Organized Prototypes), which adaptively combines complementary classification heads, regularized by organized learnable hierarchical prototypes, enabling rapid adaptation, hierarchical consistency, and structured knowledge consolidation as the taxonomy evolves. Extensive experiments on multiple benchmarks demonstrate that HALO consistently outperforms existing methods across hierarchical accuracy, mistake severity, and continual performance.

Deep Learning · Large Language Models

Ziqiang Cui, Yunpeng Weng, Xing Tang, Peiyang Liu, Shiwei Li, Bowei He, Jiamin Chen, Yansen Zhang, xiuqiang He, Rui Zhang 等

Retrieval-Augmented Generation (RAG) has emerged as a promising paradigm for improving the timeliness of knowledge updates and the factual accuracy of large language models. However, incorporating a large volume of retrieved documents significantly increases input length, leading to prohibitive computational costs. Existing compression approaches often compromise task performance, primarily due to their reliance on predefined heuristics. These heuristics fail to ensure that the compressed context is conducive to the generation tasks. To address these limitations, we propose CORE-RAG, a novel framework for context compression in RAG systems. CORE eliminates reliance on proxy heuristics through a performance-driven learning framework, which directy utilizes task performance as a feedback signal to iteratively refine the compressor policy. Prior to this optimization process, we incorporate a knowledge distillation phase to initialize the compressor with a robust policy. Extensive experiments demonstrate the superiority of our approach. At a high compression ratio of 3\%, CORE not only avoids performance degradation but also improves the average Exact Match (EM) score by 3.3 points compared to using full documents. Our code is available at https://anonymous.4open.science/r/CORE-28B4.

Deep Learning · Large Language Models

Mousa Arraf, Ido Guy, Kira Radinsky

Large language models (LLMs) often fail in systematic, model-specific ways under meaning-preserving question rewrites (paraphrases, format changes, benign distractors). In this work, we address this instability by identifying where the model's reasoning process diverges across semantically-equivalent inputs. For each target LLM, we sample multiple solution traces under rewrites and aggregate them into a graph of recurring intermediate steps, which pinpoints where incorrect traces diverge from correct ones. We then generate a small set of semantics-preserving examples that mirror the rewrite patterns most responsible for these divergences, and use them to steer the model (\emph{targeted alternation training}), either via fine-tuning or via in-context learning. Across MMLU-Pro, Big-MATH, and DROP, this yields consistent gains and cross-dataset generalization. On Humanity’s Last Exam, using 200 in-context examples, it improves GPT-5.2 (xhigh) from 35.4\% to 38.1\%, demonstrating that targeted alternation training can materially improve a frontier, API-accessible closed model under realistic access constraints.

Social Aspects · Safety

Yuhao Sun, Lingyun Yu, Hao-Xiang Xu, Fengyuan Miao, Zhuoer Xu, Hongtao Xie

Concept erasure has emerged as a promising approach to mitigate undesired or unsafe content in diffusion models, yet existing methods still face significant limitations. While training-based methods are effective, their high computational cost limits scalability. Editing-based methods are more efficient and deployment-friendly, yet they struggle to simultaneously achieve precise concept erasure and preserve overall generative capacity. We identify this core limitation of the editing-based methods as reliance on additive parameter updates. Our empirical analysis reveals that concept semantics primarily depend on *neuron direction* rather than *neuron magnitude*, while overall generative capacity relies on the *angular geometry* of neurons. As additive updates inherently entangle direction, magnitude, and angular geometry, they inevitably introduce unintended interference between concept erasure and overall generation performance. To address this, we propose **Orthogonal Concept Erasure (OCE)**, which reformulates editing-based erasure as multiplicative parameter updates from a geometric perspective. Specifically, OCE applies layer-wise orthogonal transformations derived from a closed-form solution to the parameters, enabling precise concept erasure while preserving the neuron magnitude and angular geometry. Furthermore, to address conflicting constraints in multi-concept erasure, OCE introduces a subspace-level objective with structured subspace manipulation, yielding a more effective and scalable erasure. Extensive experiments on single- and multi-concept erasure demonstrate that OCE outperforms existing methods in concept erasure and non-target preservation, erasing up to 100 concepts in 4.3 s.

Deep Learning · Everything Else

Ozan Candogan, Ayoub Foussoul

We introduce Deep Flow Networks (DFNs), a new class of discrete function approximators. DFNs are inspired by and generalize minimum-cost flow value functions that map node imbalances on a subset of nodes to the optimal flow cost. Such functions are known to be M-convex (Murota2003) and admit efficient optimization. On the theoretical side, we prove that DFNs are universal approximators for discrete functions on $\mathbb{Z}^d$ that admit convex extensions to $\mathbb{R}^d$, and characterize their optimization complexity in terms of their deviation from the M-convex regime. Guided by these results, we develop a practical DFN implementation for learning from data. Finally, we evaluate our implementation empirically on data from different ground-truth functions, showing that DFNs achieve strong approximation accuracy while being substantially faster to optimize than benchmark approaches.

Deep Learning · Graph Neural Networks

Chunyu Luo, Yuankai Luo, Xiao-Ming Wu, Lei Shi

Discrete Flow Matching (DFM) presents a promising approach for graph generation; however, existing adaptations often introduce substantial complexity by incorporating task-specific heuristics, compromising the continuity equation and significantly expanding the hyperparameter space. Moreover, their sampling efficiency remains limited, as the required number of steps is often comparable to diffusion models, diminishing DFM’s practical advantages. To address these limitations, we propose SimGFM, a simplified graph DFM for graph generation. Leveraging characteristic patterns in graph-generation trajectories, SimGFM relies only on the scheduler and rate matrix, eliminating heavy heuristics and hyperparameter tuning, and achieves large step reductions while preserving SOTA results. SimGFM achieves strong empirical results: on QM9, it matches prior models requiring 500–1000 steps with only 10 steps, and on most datasets, its performance at 50 steps matches or surpasses these baselines, demonstrating both efficiency and competitiveness.

Deep Learning · Large Language Models

Zhicheng Yang, Zhijiang Guo, Yinya Huang, Yongxin Wang, Dongchun Xie, Hanhui Li, Yiwei Wang, Xiaodan Liang, Jing Tang

Reinforcement Learning with Verifiable Reward (RLVR) is a powerful method for enhancing the reasoning abilities of Large Language Models, but its full potential is limited by a lack of exploration in two key areas: \textbf{Depth} (the difficulty of problems) and \textbf{Breadth} (the number of training instances). Our analysis of the popular GRPO algorithm reveals a bias that down-weights difficult, low-accuracy problems, which are crucial for improving reasoning skills. To address this, we introduce Difficulty Adaptive Rollout Sampling (DARS), a method that re-weights difficult problems by using targeted, multi-stage rollouts. This approach increases the number of rollout outcomes for these harder problems according to our proposed re-balancing schedules and leads to consistent gains in \textit{Pass@K}. We also found that simply enlarging the rollout size isn't effective and can even harm performance. We also investigated the role of breadth by scaling the batch size and using full-batch updates. This significantly improved \textit{Pass@1} performance by maintaining high token-level entropy, which indicates continued exploration and reduced gradient noise. Finally, we present DARS-Breadth, a combined approach that uses DARS with a large breadth of training data. This method demonstrates simultaneous gains in both \textit{Pass@K} and \textit{Pass@1}, confirming that depth (adaptive exploration) and breadth (scaling the training data) are orthogonal and essential dimensions for unlocking the full reasoning power of RLVR.

Deep Learning · Robustness

Sarthak Choudhary, Nils Palumbo, Ashish Hooda, Krishnamurthy Dvijotham, Somesh Jha

Retrieval-augmented generation (RAG) systems are vulnerable to attacks that inject poisoned passages into the retrieved context, even at low corruption rates. We show that existing attacks are not designed to be stealthy, allowing reliable detection and mitigation. We formalize a distinguishability-based security game to quantify stealth for such attacks. If a few poisoned passages control the response, they must bias the inference process more than the benign ones, inherently compromising stealth. This motivates analyzing intermediate signals of LLMs, such as attention weights, to approximate the influence of different passages on the response. Leveraging attention weights, we introduce the Normalized Passage Attention Score (NPAS) and a lightweight Attention-Variance Filter (AV Filter) that flags anomalous passages. Our method improves robustness, yielding up to ~20% higher accuracy than baseline defenses. We also develop adaptive attacks that attempt to conceal such anomalies, achieving up to 35% success rate and underscoring the challenges of achieving true stealth in poisoning RAG systems.