跳到正文
研踪PaperTrail

版本更新

查看已发布论文的最新修订,在详情页对照实际观测到的版本。

部分采集异常

全站更新概况

+1170新增论文
+937其中 AI 相关
+130已有论文修订
+94新增 AI 导读

arXiv 最近完成扫描:2026-11-19 06:59 · 全站最近入库:2026-11-19 15:04各信源进度 →

统计口径与数据新鲜度

统计 2026-11-18 17:31 — 2026-11-19 17:31(北京时间) · 点击数字查看对应简报;AI 相关包含在新增论文内。

最近完成检查:2026-11-19 06:59最近发现新增版本:2026-11-19 15:04库内论文最新更新时间:2026-11-19 02:59

每分钟刷新已保存的数据。空检查不代表采集异常;新收录可能包含历史论文。AI 导读按新增版本计数。

27,409 已收录论文22,787 AI 相关8,342 有修订112 HF 社区发现

论文修订列表

时间范围跟随排序:版本更新、本站收录或首次发表。最新收录可能包含历史论文。部分信源异常,查看状态 →

找到 8,342 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-19 06:57(北京时间)
版本更新 · v5arXiv:2606.09570

UXBench: Benchmarking User Experience in AI Assistants

Mengze Hong, Xia Zeng, Zeyang Lei, Sheng Wang, Chen Jason Zhang, Di Jiang, Taiming Fu, Jinfeng Huang 等 16 位作者

原文摘要

As AI assistants serve millions of users daily, evaluating user experience (UX) beyond general model capability has become increasingly important. We present \textbf{UXBench}, the first user-centric benchmark grounded in real user feedback signals for evaluating preference alignment and dialogue generation. The benchmark consists of three interconnected tasks, UX Judge, UX Eval, and UX Recovery, with 7,400 test instances extracted from over 70K interaction logs of a mainstream Chinese AI assistant. The dataset closely reflects real user distributions, covering 8 scenarios, 83 domains, and diverse failure patterns that pose severe challenges. Extensive experiments on 26 frontier language models provide novel insights into how well models perceive user experience and how improvements in model capability contribute to better dialogue engagement. Through analyses of model behavior and performance gaps, we document six important findings, demonstrating that user feedback prediction is a learnable capability and revealing different aspects that influence user experience. UXBench establishes a new evaluation landscape and calls for greater attention to tailored UX optimization, contributing toward a user-centric scaling law for the development of successful AI assistants. The full project is released at https://github.com/mengze-hong/UXBench.

cs.CL · cs.HC

版本更新 · v2arXiv:2510.04257

AgentTypo: Adaptive Typographic Prompt Injection Attacks against Black-box Multimodal Agents

Yanjie Li, Yiming Cao, Dong Wang, Bin Xiao

原文摘要

Multimodal agents built on large vision-language models are increasingly deployed in open-world settings but remain highly vulnerable to prompt injection. Typographic text-in-image attacks such as FigStep have been evaluated mainly on standalone vision-language models. For agents, user goals, webpage context, and action history make the effect of embedded instructions uncertain, and robustness to behavior-guided, adaptive visual injections remains insufficiently studied. We introduce AgentTypo, a black-box red-teaming framework that launches adaptive typographic prompt injection by embedding optimized text into webpage images. We formulate an attention-guided automatic typographic prompt-injection objective that favors the attacker's expected answer over the clean-page answer, together with task-conditioned visual attention from small question-answering models. A novel black-box optimization method searches text position, contrast, and color. To further strengthen the attack, we develop a multi-LLM system that iteratively refines injection prompts using evaluation feedback and retrieves successful past examples for strategy learning. Experiments on the VWA-Adv and Mind2Web benchmarks show that AgentTypo significantly outperforms state-of-the-art image-based prompt injection attacks. Across seven closed- and open-source model backends, AgentTypo-pro achieves 45%-83% ASR, outperforming the CLIP-based AgentAttack by 29-47% and previous typographic attacks by 25%-37% on VWA-Adv.

cs.CR · cs.AI

版本更新 · v2arXiv:2604.19144

ReflectMT: Internalizing Reflection for Efficient and High-Quality Machine Translation

Kunquan Li, Yingxue Zhang, Zhibin Lan, Fandong Meng, Jinsong Su

原文摘要

Recent years have witnessed growing interest in applying Large Reasoning Models (LRMs) to Machine Translation (MT). While most approaches adopt a 'pre-thinking' paradigm and benefit from explicit reasoning trajectories, they suffer from substantial inference cost and latency. To address these limitations, we propose ReflectMT, a two-stage reflection internalization framework for machine translation that employs a 'post-thinking' paradigm. Our approach develops the model's 'translate-reflect-refine' capability through reinforcement learning. In the first stage, we cultivate the model's capacity for high-quality reflection and refinement, thereby enhancing its semantic comprehension and task-specific knowledge. In the second stage, we train the model to internalize the knowledge acquired during reflection. On WMT24, ReflectMT achieves a GPT-4o-based GRF score of 86.16, outperforming DeepSeek-R1 (84.03) by 2.13 points, while using approximately 92% fewer output tokens. Independent automatic evaluation, a blinded human evaluation, and preliminary experiments on three additional translation directions provide further evidence for the framework's effectiveness.

cs.CL

版本更新 · v2arXiv:2610.30365

Data-Void Exploits in Web-Searching LLMs

Michele Guerra, Wuyeh Jobe, Sanjana Nambiar, Christina Pöpper, Azza Abouzied

原文摘要

Web-searching Large Language Models (LLMs) have changed how users seek online information: instead of sifting through ranked lists of links, users read synthesized answers. This shift creates a new LLM vulnerability: a data-void exploit. Disinformers publish their narratives in search spaces where relevant and reliable content is scarce, i.e. data voids, then rely on the LLM's web-search to retrieve and then use, cite, or list those pages in its responses to data-void seeking prompts. Unlike classical search, where users see ranking cues and competing results, LLMs can obscure when an answer relies on sparse, unverified, or adversarial sources. We study this threat by varying the user's prompt tone - skeptical, excited, or neutral - and LLM choice - ChatGPT, Gemini, or Sonar Pro - as well as three attacker controls. The attacker can use one of five known strategies to construct and fill a data void with their narrative, host their content at varying levels of visibility from well-established, indexed domains to less established ones, and include generative-optimization engine features thought to boost citing by web-searching LLMs. We find that data-void exploits can succeed with only a single attacker controlled page. Across all conditions, attacker-content appears in LLM responses 31% of the time. User controls offer limited protection: while skeptical prompts slightly reduce success and slightly increase flagging or warnings in LLM responses, we find no evidence of large effects of prompt tone or model choice on success rates, although smaller effects cannot be ruled out. Overall, our results establish data voids as a practical inference-time attack surface for manipulating web-searching LLMs through strategically placed web content and highlight the urgent need for defenses against such exploits.

cs.CR · cs.HC · cs.IR

版本更新 · v2arXiv:2610.05197

From Scientific Observations to Mechanisms: Benchmarking Hypothesis Generation by AI Scientists

Xiaxun Xie, Qingqing Long, Meng Xiao, Wei Ju, Yuanchun Zhou, Hengshu Zhu, Xuezhi Wang

原文摘要

Data-driven mechanistic hypotheses are essential to scientific discovery because they explain how underlying processes produce observed phenomena. AI agents and AI scientists increasingly support scientific data analysis. However, their ability to turn empirical findings into mechanistic hypotheses remains insufficiently examined. To address this gap, we introduce MechHypoBench, the first benchmark for evaluating whether AI agents and AI scientists can generate such hypotheses from empirical data. It combines paper-derived mechanisms from 14 scientific fields with real-world datasets containing 17.98 million records. The construction retains the observational complexity of empirical data while providing a specified underlying mechanism. Agents analyze the observations and propose open-form hypotheses. We develop an evaluation framework that assesses open-form mechanistic hypotheses through their consequences under withheld conditions. Experiments with general agents and AI scientists reveal a substantial gap between generated hypotheses and the underlying mechanisms.

cs.AI

版本更新 · v3arXiv:2605.24202

When Does Multi-Agent RL Improve LLM Workflows? Workflow, Scale, and Policy-Sharing Tradeoffs

Yifan Zeng, Yiran Wu, Yaolun Zhang, Wentian Zhao, Kun Wan, Qingyun Wu, Huazheng Wang

原文摘要

Outcome-reward RL can train every role of a multi-agent LLM workflow. Its benefit is hard to predict because the reward assigns no credit to individual roles and a shared policy trades pooled data against specialization. We ask when this training improves a workflow over its base model and single-agent RL, and what each role learns. We train three workflows (Eval-Opt, Voting, Orch-Workers) with GRPO on math and code at three Qwen3 scales, with one policy for all roles (Shared-Policy, SP) or one per role type (Isolated-Policy, IP). Every workflow gains by adapting to its structure; this does not make a role better than training it alone. Single-agent RL in the same structure matches or beats trained Voting and Orch-Workers, and within the grid only Eval-Opt under IP exceeds it. Eval-Opt's evaluator learns when to retry and its generator learns to complete the retried attempts, a joint strategy that a single policy trained alone does not acquire. Under SP, per-output token averaging sets each role's gradient share, a hidden hyperparameter absent from single-agent RL, and the solving roles learn slowly; reweighting it gives SP the speed and peak of IP, and its decline. Under IP, a parallel role, whose instances share one adapter, takes one optimizer step per instance per training step and can drift after the peak as its entropy rises and accuracy declines. Merging those steps delays the decline. Our code is at: https://github.com/XHMY/marl-llm-workflows

cs.AI · cs.LG

版本更新 · v2arXiv:2607.15880

Dynamics-Aware Meta-Imitation for Generalization to Unseen Robotic Manipulation

Zhenduo Shang, Xiyao Liu, Bohan Li, Xudong Wang, Teng Ren, Lianqing Liu, Zhi Han

原文摘要

Imitation Learning aims to learn skills from extensive observations and demonstrations for robots, so it suffers from data scarcity and environment generalization. The existing methods predominantly focus on imitation from in-domain tasks and consequently struggle with generalization to unseen tasks. To bridge this generalization gap, we propose the \textbf{D}ynamics-\textbf{A}ware \textbf{M}eta-\textbf{I}mitation (DAMI) framework. By integrating meta-learning to construct a shared skill space, DAMI equips agents for rapid adaptation to novel tasks. We introduce the Visual-Motor Trajectory (VMT) module to capture complex spatio-temporal dynamics within the task latent space. Furthermore, we propose the Unpaired Unified Task (U2T) block to fuse unstructured multimodal observations. To coordinate these representations, we integrate a Task-Conditioned Feature Modulation (TCFM) mechanism customized for modulating low-level 3D features. By capturing intrinsic dynamics from a random complete reference demonstration, our framework learns the underlying task logic rather than memorizing static cues, ensuring effective generalization. Extensive experiments in both simulation and real-world settings demonstrate that our approach outperforms state-of-the-art baselines regarding direct inference on seen tasks and adaptation to unseen tasks via few-shot fine-tuning.

cs.RO · cs.LG

版本更新 · v2arXiv:2601.12428

ReWorld: Multi-Dimensional Reward Modeling for Embodied Video Generation

Baorui Peng, Wenyao Zhang, Liang Xu, Bohan Li, Hu Zhu, Zekun Qi, Hongsi Liu, Jiazhao Zhang 等 10 位作者

原文摘要

Current embodied video generation models primarily emphasize visual generative quality while overlooking physical fidelity, dynamic consistency, and task logic, especially for contact-rich manipulation tasks, which limits their applicability to downstream tasks. In this paper, we introduce ReWorld, a framework aimed to employ reinforcement learning to align the video generation with physical realism, task completion capability, embodiment plausibility and visual quality. Specifically, we first construct a large-scale (~235K) video preference dataset and employ it to train a hierarchical reward model designed to capture multi-dimensional reward consistent with human preferences. We further propose a practical alignment algorithm that post-trains flow-based video generators using this reward through a computationally efficient PPO-style algorithm. Comprehensive experiments and theoretical analysis demonstrate that ReWorld significantly improves the physical fidelity, logical coherence, embodiment and visual quality of generated rollouts, outperforming previous methods.

cs.RO · cs.CV

版本更新 · v3arXiv:2603.17450

VLM2Rec: Resolving Modality Collapse in Vision-Language Models for Sequential Recommendation

Junyoung Kim, Woojoo Kim, Jaehyung Lim, Dongha Kim, Wonbin Kweon, Hwanjo Yu

原文摘要

Sequential Recommendation (SR) learns sequence-level Collaborative Filtering (CF) from interactions, but ID-based models struggle with sparse data and cold-start items. In contrast, item titles and images can reveal preference continuity, such as a recurring color or style. Yet existing approaches miss visual continuity: small encoders take a single item as input, and Large Language Models (LLMs) read only titles. We therefore shift our focus to Large Vision-Language Models (VLMs), which can read a user's entire image sequence, not just its titles. However, VLMs inherently suffer from modality collapse: they over-rely on text and, as a result, overlook visual continuity for recommendation. We find that standard contrastive Supervised Fine-Tuning (SFT), essential for adapting VLMs to SR, further amplifies this collapse: updates follow the text gradient, visual discrimination weakens, and adding image embeddings to the text embeddings consistently lowers accuracy. We propose VLM2Rec with two objectives: Weak-modality Penalized Contrastive Learning strengthens each user's weaker modality, and Cross-modal Relational Topology Regularization aligns each user's relative similarity structure over the same items across the two modality spaces. Its variant, $\text{VLM2Rec}_{\text{plug}}$, reuses the learned item embeddings in SR backbones and thus serves at the latency of conventional SR models. Experiments show that VLM2Rec mitigates the collapse and consistently improves accuracy and robustness.

cs.IR · cs.AI

版本更新 · v2arXiv:2609.11308

2AM: Grounding Agent-Side Memory as Guidance for Steerable Action Models in Long-Horizon Manipulation

Yutong Hu, Fengjiao Chen, Xuezhi Cao, Renaud Detry

原文摘要

Long-horizon robot manipulation requires memory, but not necessarily inside the action policy. To address such tasks, current agentic systems often combine VLAs with planners and geometric tools, sometimes using additional depth or calibrated geometry. These systems confound attribution: gains may come from richer observations or alternative motor tools, while failures may stem from either the policy or an under-specified language interface. We isolate this question through a deliberately constrained design: less tool breadth, but greater interface bandwidth. 2AM makes a multimodal Agent the sole holder of task memory and a single RGB-based, episodically stateless Action Model the sole executor of task-relevant motion. The Agent compiles interaction history into subtask language and optional 2D grasp, place, and move hints that bind its physical intention at different time scales. To teach this steerability to the VLA, we augment demonstrations with structured hint labels and train under condition dropout, spatial noise, and temporal jitter to tolerate imperfect Agent outputs. On LIBERO-Mem, without depth, online geometry, or planner-based object motion, 2AM reaches 76.3% average completion, a 61.5-point improvement over the strongest reported baseline of 14.8%, together with 63.0% relaxed and 11.8% strict success. These results show that task memory can remain Agent-side. They further show that Action Model capability depends not only on what the policy has learned, but on how precisely the Agent can steer it. Project website: agent-mem-action-model.github.io

cs.RO · cs.AI

版本更新 · v2arXiv:2605.24044

MERIS: Adaptive Real-Time Scheduling of Refined Inference DAGs in Dynamic Robotic Environments

Zexin Li, Tao Ren, Johnathan Liu, Xiaoxi He, Cong Liu

原文摘要

Robots deployed in dynamic environments execute changing directed-acyclic-graph (DAG) workloads under limited compute and timing budgets. This article presents MERIS, a framework for model scheduling under real-time constraints that combines proportional intermediate deadlines, MIMONet-aware graph refinement, and event-driven synchronization for multi-task inference on resource-constrained robotic platforms. We distinguish three representations: the application DAG, each MIMONet's internal encoder/decoder graph, and the refined scheduling DAG, and make the scope of the timing argument explicit. If the profiled execution bounds, release assumptions, precedence barriers, and per-level capacity premises hold, satisfaction of the assigned sub-deadlines composes to the end-to-end deadline. For non-partitionable tasks, we define a profile-bounded admission rule; before interpreting that rule as empirical timing evidence, we require direct per-task completion checks. We report run-level QoE alongside the offered-node deadline missing rate, making explicit that run completion and successful service of every node are different outcomes. We propose a comprehensive minibenchmark for evaluating the performance of schedulers under various DNN-driven autonomous driving scenarios. Extensive experiments on Nano Orin, AGX Orin, and M3 Max demonstrate the effectiveness of our proposed method.

cs.RO · cs.SE · eess.SY

版本更新 · v2arXiv:2603.03280Berkeley BAIR

How to Peel with a Knife: Aligning Fine-Grained Manipulation with Human Preference

Toru Lin, Shuying Deng, Zhao-Heng Yin, Pieter Abbeel, Jitendra Malik

原文摘要

Many essential manipulation tasks - such as food preparation, surgery, and craftsmanship - remain difficult for autonomous robots. These tasks are characterized not only by contact-rich, force-sensitive dynamics, but also by their "implicit" success criteria: unlike pick-and-place, task quality in these domains is continuous and subjective (e.g. how well a potato is peeled), making quantitative evaluation and reward design difficult. We present a learning framework for such tasks, using peeling with a knife as a representative example. Our approach follows a two-stage pipeline: first, we learn a robust initial policy via force-aware data collection and imitation learning, enabling generalization across object variations; second, we refine the policy through preference-based finetuning using a learned reward model that combines quantitative task metrics with qualitative human feedback, aligning policy behavior with human-evaluated task quality. Using only 50~200 peeling trajectories, our system achieves 100% same-produce success on cucumbers, apples, and potatoes, with performance improving by up to 40% through preference-based finetuning. Remarkably, policies trained on one produce category also transfer zero-shot to selected held-out produce categories, achieving 50% success on zucchini, 90% on pear, and 80% on daikon radish.

cs.RO · cs.AI · cs.CV · cs.LG · eess.SY

版本更新 · v2arXiv:2610.05264

Task-Aware Joint Pruning and Distillation for Efficient Audio Deepfake Detection

Miao He, Peng Cheng, Zhongjie Ba, Qing Wen, Li Lu, Xin Yang, Kui Ren

原文摘要

Advances in speech synthesis have made deepfake speeches increasingly convincing, posing growing threats to security. While self-supervised learning (SSL) based detectors achieve state-of-the-art performance, their computational demands (typically 300M+ parameters) prevent deployment on resource-constrained devices. Existing compression methods, designed mainly for content-centric tasks, struggle to maintain competitive performance when directly adapted to deepfake detection. We propose a Task-Aware Joint Pruning and Distillation framework that combines cross-domain knowledge distillation with movement-guided structured pruning to transfer forgery-discriminative knowledge and preserve critical structures under aggressive compression. Our framework reduces the model to 31.9M parameters with 6.3$\times$ FLOPs reduction, with an average performance drop of only 1.30\% across multiple datasets compared to the uncompressed baseline, demonstrating strong potential for on-device deployment.

cs.SD · cs.AI

版本更新 · v6arXiv:2605.19398

Rebalancing Reference Frame Dominance to Improve Motion in Image-to-Video Models

Wooseok Jeon, Seungho Park, Seunghyun Shin, Sangeyl Lee, Hyeonho Jeong, Hae-Gon Jeon

原文摘要

Image-to-video models often generate videos that remain overly static, compared to text-to-video models. While prior approaches mitigate this issue by weakening or modifying the image-conditioning signal, they often require additional training or sacrifice fidelity to the reference image. In this work, we identify reference-frame dominance as a key mechanism behind motion suppression. We observe that non-reference frames in I2V models allocate excessive self-attention to reference-frame key tokens, causing reference information to be over-propagated across time and suppressing inter-frame dynamics. Based on this finding, we propose DyMoS (Dynamic Motion Slider), a training-free and model-agnostic method that rebalances the attention pathway from generated frames to the reference frame during initial denoising steps. DyMoS leaves both the input image and model weights unchanged and introduces a single scalar parameter for continuous control over motion strength. Experiments across multiple state-of-the-art I2V backbones demonstrate that DyMoS consistently improves motion dynamics while maintaining visual quality and fidelity to the reference image.

cs.CV · cs.AI

版本更新 · v4arXiv:2602.02220

LangMap: A Human-Verified Benchmark for Hierarchical Open-Vocabulary Goal Navigation

Bo Miao, Weijia Liu, Jun Luo, Lachlan Shinnick, Jian Liu, Thomas Hamilton-Smith, Yuhe Yang, Zijie Wu 等 11 位作者

原文摘要

Language-conditioned goal navigation requires embodied agents to locate user-specified targets without step-by-step guidance. However, existing benchmarks largely focus on category-level goals or rely on instance descriptions generated by vision-language models, which often contain ambiguities and semantic errors, limiting systematic and reliable evaluation. We introduce HieraNav, an open-vocabulary navigation task with goals at four hierarchical semantic levels: scene, room, region, and instance. We present Language as a Map (LangMap), the first goal navigation benchmark to enrich real-world indoor 3D scans with human-verified semantic annotations supporting tasks across all four goal levels. Built on HM3D using a contrastive annotation protocol that compares same-scene regions and instances, LangMap provides region labels and discriminative region and instance descriptions covering 414 object categories and contains over 18K tasks. Each target has concise and detailed descriptions, enabling evaluation across instruction styles. Automated and human evaluations validate our annotation quality: our descriptions improve text-to-view matching accuracy over GOAT-Bench's by 23 points on all shared annotated instances, and an independent human audit yields 92.5% unique-and-correct matches. We also propose PlaNaVid, an RGB-only baseline that combines Bounded Diverse Memory with high-level planning to prime a reactive policy for multi-goal navigation, achieving top-tier success rates without depth, 3D scene representations, or object masks. Further analyses reveal that exploration and hierarchical disambiguation failures become more prominent at finer goal levels, while long-tail categories, small objects, distant targets, timely stopping, and multi-goal completion remain challenging. Benchmark and code: https://bo-miao.github.io/LangMap

cs.CV · cs.RO

版本更新 · v2arXiv:2609.04822

Minimax Rate for Diffusion-Guided Local Intrinsic Dimension Estimation based on Fokker-Planck Equation

Jaehee Seo, Wontae Jeong, Jisu Kim

原文摘要

Diffusion-guided methods have recently emerged as effective tools for probing the intrinsic geometry of high-dimensional data. Understanding their accuracy requires quantifying both (i) finite-scale approximation error relative to the true local intrinsic dimension (LID) and (ii) estimation error from finite samples. We address these two questions for the FLIPD field (Kamkari et al., 2024; arXiv:2406.03537), defined at the population level through the logarithmic scale derivative of a Gaussian-smoothed density and admitting a Fokker--Planck representation. For $α$-smooth densities on a regular $d$-dimensional manifold, we establish a uniform $O(σ^2)$ approximation bound relative to the manifold dimension $d$. Under expected integrated squared loss, we prove a minimax lower bound of order $(nσ^d)^{-1}$ for estimating the FLIPD field from $n$ observations when $n^{-1/(2α+d)}\leσ\leσ_0$. A plug-in FLIPD estimator based on the empirical denoising score matching (DSM) solution attains a matching upper bound. Thus the minimax rate is of order $(nσ^d)^{-1}$ and it becomes $n^{-2α/(2α+d)}$ at its smallest scale.

stat.ML · cs.LG · math.ST

版本更新 · v2arXiv:2610.39884

KARMA: Training-Free Knowledge-Atomic Harness Evolution for Industrial Advertising Agents

Xinhua Miao, Yuanzhen Rong, Weiru Zhang, Xiaoyi Zeng

原文摘要

Industrial advertising agents must continually adapt their decision knowledge to changing product offerings, traffic patterns, and operating policies. We present KARMA (Knowledge-Atomic, Regression-Gated Modular Adaptation), a training-free framework that evolves six typed knowledge modules while keeping model weights and executable harness code frozen. During our evolution, a deterministic Trace Verifier maps observed errors to candidate modules, and a Critic compares successful and failed trajectories to guide a tool-constrained Evolver. The Evolver applies atomic knowledge edits, which a fixed validation gate retains only when their measured gain reaches a threshold; otherwise, the previous knowledge state is restored. We evaluate KARMA with three LLM families on cases sampled from a corpus of approximately half a million real-world operation logs across six e-commerce domains. Across 18 domain--model settings, KARMA raises the average hold-out score from 30.19 to 39.15 (+8.96 points, +29.7%). Ablations show that removing either successful or failed trajectories more than halves the average hold-out gain. Evolved knowledge also transfers across domains without further adaptation, yielding an average hold-out gain of 8.54 points across all source--target pairs. By linking retained patches to their diagnostic evidence and validation outcomes, KARMA supports inspectable, reversible knowledge adaptation for agents in Web marketplaces.

cs.AI

版本更新 · v5arXiv:2508.04903

RCR-Router: Efficient Role-Aware Context Routing for Multi-Agent LLM Systems with Structured Memory

Jun Liu, Zhenglun Kong, Changdi Yang, Fan Yang, Tianqi Li, Peiyan Dong, Joannah Nanjekye, Hao Tang 等 15 位作者

原文摘要

Large language model (LLM) agents are increasingly organized into multi-agent systems for complex reasoning and decision-making tasks. However, scaling such systems can substantially increase inference cost, especially when multiple agents repeatedly process overlapping or role-irrelevant context. Many existing multi-agent frameworks focus on orchestration and role specialization, but still rely on full-context sharing or static context routing, leading to redundant token consumption, increased runtime, and inefficient memory utilization. We present \emph{Role-Conditioned Context Routing (RCR)}, an execution-time layer that allocates role-relevant shared-memory items to agents under token budgets. Given a structured shared memory and a set of role-specialized agents, RCR selectively routes task-relevant memory items to each agent according to its role, optional task-stage metadata, and token budget. RCR integrates role-conditioned relevance scoring, token-budgeted memory selection, and iterative memory updates to reduce redundant context exposure during multi-agent execution. Experiments on multi-hop QA, embodied reasoning (ALFWorld), and web-based decision-making (WebShop) show that RCR reduces token usage by 35\% on average and runtime by 38\% on average relative to Full-Context, while maintaining or improving task performance.

cs.CL · cs.AI · cs.MA

版本更新 · v2arXiv:2606.19168

Beyond Safe Data: Pretraining-Stage Alignment with Regular Safety Reflection

Jinhan Li, Kexian Tang, Yihan Xu, Zhuorui Ye, Kaifeng Lyu

原文摘要

To achieve deeper safety alignment for large language models (LLMs), recent efforts have studied how to push safety interventions earlier into the pretraining stage, primarily by filtering unsafe data or rewriting it into safer forms. We argue that pretraining-stage alignment should go beyond making the data safe: LLMs may generalize unsafe behaviors from seemingly benign pretraining data. To this end, we propose Safety Reflection Pretraining, a pretraining-stage alignment method which regularly inserts short safety reflections between sentences in pretraining corpora. This integrates self-monitoring directly into language modeling, establishing a foundational capability that is subsequently reinforced by compatible post-training. Our experiments with 1.7B models pretrained on FineWeb-Edu show that Safety Reflection Pretraining improves safety classification accuracy and substantially reduces the success rates of inference-stage and fine-tuning attacks. Complementary to our real-world experiments, we also introduce a fully controlled synthetic environment, MedSafetyWorld, with a clear definition of safety and a reasoning structure under which we find that models can easily generalize unsafe behaviors from safe data. Ablations in MedSafetyWorld further demonstrate a clear advantage of Safety Reflection Pretraining over data filtering and rewriting in preventing models from acting on unsafe behaviors. Taken together, our findings suggest that pretraining alignment should not only make the training data safe, but also shape the behaviors that models are likely to acquire from safe data.

cs.AI · cs.LG

版本更新 · v3arXiv:2507.16221

Unbeatable imitation of a friend

Masahiko Ueda

原文摘要

Imitation sometimes achieves success in multi-agent situations even though it is very simple. In repeated games, success of imitation has been characterized by unbeatability against other agents. Previous studies specified conditions under which imitation is unbeatable against co-players in repeated symmetric games, and clarified that the existence of unbeatable imitation is strongly related to the existence of payoff-controlling strategies, called zero-determinant strategies. However, these results were on the situations in which one repeated game is played by one group of players. It was pointed out that imitation of other players in the same group and imitation of other players in the same role in other groups generally result in different outcomes. Here, we investigate the existence condition of unbeatable imitation in the latter situations, described by two concurrent repeated games played by two different groups of players. We find that the existence condition of unbeatable imitation is stronger than the existence condition of unbeatable zero-determinant strategies, whereas both are very limited. Furthermore, we find that adding exploration can make an imitation strategy unbeatable even in games where imitation without exploration can be beaten. Our findings suggest a strong relation between unbeatable imitation and payoff control even in the two concurrent repeated game situations.

physics.soc-ph · cs.MA · eess.SY

版本更新 · v2arXiv:2610.41365

Generated storm lifecycles fill the rare-event tail of short rainfall records

Valerie Tsao, Nathaniel W. Chaney, Manolis Veveakis, Marta Zaniolo

原文摘要

Historically rare ''one-in-a-generation'' storms are occurring with unprecedented frequency, causing catastrophic damage and mounting casualties, and exposing the vulnerability of infrastructure designed around short observational records. To overcome the limitations of sparse data, we present StormGen, a generative machine-learning model that transforms an observed storm's initial hours into complete, physically plausible hourly rainfall lifecycles in minutes. Evaluated on a decade of southeast Texas rainfall, StormGen produced close counterparts for 4\5% of the most severe 2\% of storms--compared to just 3\% for stochastic storm transposition--and maintained spatial coherence over full 24-hour sequences. Crucially, from a single hour of initial rainfall, the model successfully reconstructed the specific lifecycle, spatial pattern, and extreme catchment totals of Hurricane Harvey itself, despite the event being entirely absent in training. Because storm timing alone can more than double peak flows, these complete generated scenarios provide a vital mechanism to stress-test flood protections against unprecedented extremes.

cs.LG

版本更新 · v3arXiv:2605.19232

Devilray: A Reconfigurable Adversarial Baseline for Revealing Blind Spots in Fake Base Station Detection

Taekkyung Oh, Duckwoo Kim, Hansung Bae, Beomseok Oh, CheolJun Park, Tyler Tucker, Nathaniel Bennett, Sangwook Bae 等 11 位作者

原文摘要

Fake Base Station (FBS) detection has been a critical focus of cellular security research for over two decades. However, significant financial and regulatory barriers to accessing commercial FBS (C-FBS) devices have limited direct visibility into real-world operations, forcing detection systems to be designed and evaluated around self-built prototypes. In this paper, we present Devilray, a reconfigurable adversarial baseline to guide effective FBS detection by exploring the realistic adversarial space and identifying blind spots in detection systems -- regions of realistic adversarial behavior excluded by prevailing threat models. To this end, we establish an empirical anchor through the first academic analysis of a C-FBS, define a fundamental four-phase pipeline and seven operational feature dimensions for adversarial behaviors, and extend them with operational variants derived from 3GPP standards analysis. Through combinatorial configurations of operations from commercial practices and specifications, Devilray enables the exploration of 2,592 FBS instances, spanning diverse operational possibilities. Using Devilray, we evaluate seven currently accessible FBS detectors and uncover coverage gaps across all seven, revealing blind spots rooted in assumption-bound design and evaluation. Our work provides an empirically grounded adversarial baseline for evaluating FBS detectors and guiding future detection design, with Devilray and its configuration profiles released to vetted research groups.

cs.CR

版本更新 · v5arXiv:2609.24965

Jev for Scientific Decisions: Evaluating Semantic Choices and Their Consequences

Boyuan Deng, Shuyi Fan, Hongyang Zhang, Xinhong Xie

原文摘要

Scientific workflows often require choosing among known relations before a deterministic calculation can proceed. Whether observations share a culture, treatment or reference standard can change the scientific meaning of the resulting count or comparison. Once the evidence is prepared and the calculation is written as code, this choice reduces to one semantic judgment among listed candidates. We evaluate Jev on this step, using a harness that follows its documented guidance. The study compares twelve model configurations on twenty Choices in ten scientific cases built from published sources, each repeated five times. We measure semantic selections, downstream outputs and final claim labels separately. Jev tied with five other configurations at complete semantic correctness and achieved the lowest observed median latency and known mean cost among successful responses. On one question, whether samples shared a culture, three comparison models made seven wrong selections that changed downstream counts while preserving the correct final label. These results support using Jev to select relations when the evidence and candidates are prepared in advance, and show that evaluating it requires checking the relations and quantities that a workflow will reuse.

cs.CL · cs.AI

版本更新 · v5arXiv:2507.08846

Precomputed Dominant Resource Fairness

Serdar Metin

原文摘要

Dominant Resource Fairness (DRF) is a resource allocation scheme, addressing the problem of fair allocation of multiple resource types among users with heterogeneous demands for different types of resources. Based on Max-min Fairness, which is a well-established algorithm in the literature for allocating resources in single resource type scenarios, DRF generalises the main principle of maximising the minimum share given to any user to the multiple resource case. It has a number of desirable properties, such as Sharing Incentive, Envy-Freeness, Pareto Efficiency, and Strategy Proofness, that make it preferable over alternatives, and as such, it is widely adopted in distributed systems. In the present study, we revisit the original study and analyse the structure of the algorithm in closer view. The main contribution is an alternative algorithm, which approximates the DRF allocation in linear time, significantly improving over the original algorithm's pseudo-polynomial runtime. We name the new algorithm Precomputed Dominant Resource Fairness (PDRF), after its main working principle. We provide bounds for approximating the original algorithm, and critically evaluate the agreements of aforementioned properties, comparing DRF and PDRF over them. It turns out that there is an adversarial strategy which forces DRF to unfairly allocate more resources to users who submit inflated demands. PDRF, on the other hand, favouring smaller demands, penalises such adversarial behaviour and creates incentive for users to submit their minimal demands.

cs.GT · cs.DC · cs.DS

版本更新 · v3arXiv:2502.13828

Software Security in Software-Defined Networking: A Systematic Literature Review

Moustapha Awwalou Diouf, Samuel Ouya, Jacques Klein, Tegawendé F. Bissyandé

原文摘要

Software-defined networking (SDN) separates the control plane from the data plane and exposes the network through software applications and open APIs. The same programmability that drove its adoption also turned the network into a large body of software, and that software can be vulnerable. Knowing where and how SDN software fails is a prerequisite for defending it, yet the literature offers no consolidated account of the problem. We address this gap with a systematic literature review of 113 primary studies published between 2012 and 2025 on the security of the software that makes up SDN. We study how each software component becomes vulnerable, how attackers exploit it, which testing and analysis techniques expose its defects, and how the field has evolved. The review yields a taxonomy of vulnerabilities and attack vectors organized by SDN software component, a synthesis of the methods used to find software defects in each component, and a set of open problems that mark the most promising directions for future work. Earlier surveys treat SDN security as a networking problem; ours is, to our knowledge, the first to treat SDN components as software artifacts whose code, logic, and interactions can be defective. Our artifacts are available at https://github.com/mad975/SDNSoftwareSecurity_SLR.

cs.CR · cs.SE

版本更新 · v4arXiv:2503.18974

An Efficient Frequency-Based Approach for Maximal Square Detection in Binary Matrices

Swastik Bhandari

原文摘要

We present FreqBased, a frequency-based algorithm for the maximal square problem in binary matrices. The algorithm maintains a vertical run-length array and a horizontal counter to validate candidate squares, matching the standard single-array dynamic programming algorithm in O(mn) time and O(n) space while reducing the amount of work performed during execution. Experiments on synthetic random binary matrices reveal a clear density-dependent performance transition: FreqBased consistently outperforms the dynamic programming baseline on moderate- and high-density inputs across Intel and AMD processors, achieving peak speedups of 1.175x and 1.277x, respectively, while executing up to 2.08x fewer instructions. On 180 real aerial image masks from the INRIA dataset, FreqBased outperforms the baseline on every instance with speedups of up to 4.63x, confirming that the instruction-count advantage observed on synthetic inputs extends to real binary matrices with spatial structure. Hardware performance counter analysis shows that the performance advantage arises primarily from reduced instruction count despite lower execution efficiency, demonstrating that algorithms with identical asymptotic complexity can exhibit substantially different practical performance depending on input characteristics.

cs.DS · math.OC

版本更新 · v2arXiv:2503.21049

On the Hardness Hierarchy for the $O(n \sqrt{\log n})$ Complexity in the Word RAM

Dominik Kempa, Tomasz Kociumaka

原文摘要

We study the relative hardness of fundamental problems with state-of-the-art word RAM algorithms that take $O(n\sqrt{\log n})$ time for instances described in $Θ(n)$ machine words ($Θ(n\log n)$ bits). This complexity class, one of six hardness levels identified by Chan and Pătraşcu [SODA 2010], includes problems from several domains: Counting Inversions, string processing problems (BWT Construction, LZ77 Factorization, Longest Common Substring, Batched Longest Previous Factor Queries, Batched Inverse Suffix Array Queries), and computational geometry tasks (Orthogonal Range Counting, Orthogonal Segment Intersection Counting). We offer two main contributions: We establish new links between the above string problems and Dictionary Matching, a classic task solvable using the Aho-Corasick automaton. We restrict Dictionary Matching to instances with $Θ(n)$ binary patterns of length $m = Θ(\log n)$ each, and we prove that, unless these instances can be solved in $o(n\sqrt{\log n})$ time, the aforementioned string problems cannot be solved faster either. Via further reductions, we extend this hardness to Counting Inversions (a fundamental component in geometric algorithms) and thus to Orthogonal Range Counting and Orthogonal Segment Intersection Counting. This hinges on String Nesting, a new problem which is equivalent to Dictionary Matching and can be reduced to Counting Inversions in three steps. Together, our results unveil a single problem, with two equivalent formulations, that underlies the hardness of nearly all major problems currently occupying the $O(n\sqrt{\log n})$ level of hardness. These results drastically funnel further efforts to improve the complexity of near-linear problems. As an auxiliary outcome of our framework, we also prove that several central string problems can be efficiently reduced to their versions over the binary alphabet.

cs.DS

版本更新 · v2arXiv:2610.37682

Rank-Aware Differentially Private Release of Listwise Preferences for LLM Alignment

Junwei Chen, Manjiang Yu, Pengpeng Qiao, Yang Cao

原文摘要

Listwise preference feedback offers richer supervision for large language model (LLMs) alignment than pairwise comparisons, but a full ranking also reveals more sensitive user preferences. Existing privacy-preserving alignment methods focus mainly on pairwise feedback, while listwise alignment under local differential privacy (LDP) remains unexplored. Extending LDP to this listwise domain raises a critical challenge: the high sensitivity of the full ranking space demands excessive noise, neutralizing the benefits of listwise supervision. To address this challenge, we analyze the interaction between supervision granularity and privacy noise in downstream training, and introduce a rank-aware exponential mechanism that privatizes listwise preference data into a low-sensitivity granularity suitable for downstream alignment. The mechanism leverages ranking information to sample a fixed-size binary partition, concentrating the noise near borderline items instead of perturbing all items uniformly as randomized response (RR) does. Empirical evaluations on downstream LLM alignment tasks show that our mechanism consistently outperforms existing LDP baselines at matched privacy budgets. The code is available at https://github.com/tdsai-lab/RAEM.

cs.CL · cs.AI

版本更新 · v5arXiv:2603.06082

Offline Materials Optimization with CliqueFlowmer

Jakub Grudzien Kuba, Benjamin Kurt Miller, Sergey Levine, Pieter Abbeel

原文摘要

Recent advances in deep learning inspired neural network-based approaches to computational materials discovery (CMD). A plethora of problems in this field involve finding materials that optimize a target property. Nevertheless, the increasingly popular generative modeling methods are ineffective at boldly exploring attractive regions of the materials space due to their maximum likelihood training. In this work, we offer an alternative CMD technique based on offline model-based optimization (MBO) that fuses direct optimization of a target material property into generation. To that end, we introduce a domain-specific model, dubbed CliqueFlowmer, that incorporates recent advances of clique-based MBO into transformer and flow generation. We validate this model's optimization abilities and show that materials it produces strongly outperform those from generative baselines. To support specialized materials discovery applications and broader interdisciplinary research, we release our code (https://github.com/znowu/CliqueFlowmer), model weights (https://huggingface.co/iamkuba/CliqueFlowmer), and an interactive demo (https://cliqueflowmer.znowu.workers.dev).

cs.AI · cs.CE

版本更新 · v3arXiv:2610.11213

AliO: Output Alignment Matters in Long-Term Time Series Forecasing

Kwangryeol Park, Jaeho Kim, Seulki Lee

原文摘要

Long-term Time Series Forecasting (LTSF) tasks, which leverage the current data sequence as input to predict the future sequence, have become increasingly crucial in real-world applications such as weather forecasting and planning of electricity consumption. However, state-of-the-art LTSF models often fail to achieve prediction output alignment for the same timestamps across lagged input sequences. Instead, these models exhibit low output alignment, resulting in fluctuation in prediction outputs for the same timestamps, undermining the model's reliability. To address this, we propose AliO (Align Outputs), a novel approach designed to improve the output alignment of LTSF models by reducing the discrepancies between prediction outputs for the same timestamps in both the time and frequency domains. To measure output alignment, we introduce a new metric, TAM (Time Alignment Metric), which quantifies the alignment between prediction outputs, whereas existing metrics such as MSE only capture the distance between prediction outputs and ground truths. Experimental results show that AliO effectively improves the output alignment, i.e., up to 58.2% in TAM, while maintaining or enhancing the forecasting performance (up to 27.5%). This improved output alignment increases the reliability of the LTSF models, making them more applicable in real-world scenarios.

cs.AI