跳到正文
研踪PaperTrail

论文库

检索计算机领域论文,按研究方向、作者或关键词筛选。

部分采集异常
正在读取账号与研究记录…

全站更新概况

+197新增论文
+151其中 AI 相关
+16已有论文修订
+95新增 AI 导读

arXiv 最近完成扫描:2026-11-21 12:19 · 全站最近入库:2026-11-21 12:19各信源进度 →

统计口径与数据新鲜度

统计 2026-11-20 23:02 — 2026-11-21 23:02(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:2026-11-21 12:19最近发现新增版本:2026-11-21 12:19库内论文最新更新时间:2026-11-20 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。

29,158 已收录论文24,181 AI 相关8,992 有修订112 HF 社区发现

浏览论文库

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 138 篇论文 · 按版本更新时间排序上轮扫描起点 2026-11-21 12:17(北京时间)
版本更新 · v2arXiv:2611.14961

CommJEPA: When Predictive Representation Learning Meets Wireless Intelligence

Feiran You, Hongyang Du, Kang Wei, Jun Li, Abbas Jamalipour

原文摘要

Artificial intelligence (AI) is becoming an integral part of wireless systems, yet much of today's wireless AI remains task-specific and reactive, with models typically trained to map current observations to predefined outputs or reconstruct observed data. While effective for individual tasks, such approaches provide limited support for learning what is predictable, reusable, and transferable in dynamic wireless environments. Joint-Embedding Predictive Architecture (JEPA) offers a different perspective by learning predictive representations in latent space that capture informative structure in the observations. In this article, we introduce Communication JEPA (CommJEPA) as a communication-oriented framework for predictive representation learning. The central idea is bidirectional. JEPA can equip wireless systems with predictive representations for capturing latent channel and environmental dynamics, integrating heterogeneous observations, adapting to changing conditions, and supporting future decisions. Conversely, the physical structure, partial observability, multiple time scales, action-dependent evolution, and resource constraints of wireless systems motivate new communication-aware JEPA designs. We develop this mutual perspective, organize the resulting design space, and discuss representative JEPA-based wireless instantiations for prediction and control. We further highlight open challenges toward scalable, interactive, and predictive wireless intelligence.

cs.NI

新论文 · v1arXiv:2611.25695

DMT-Attack: Diffusion-Prior Texture Search for Closed-Loop Vision-Language-Action Policies

Jun Xia, Shangwen Wang, Jun Ma, Ruichao Ren, Zhouyang Jia, Xin Zhao, Huijun Liu, Jie Yu

原文摘要

Vision-language-action (VLA) policies rely on visual feedback throughout manipulation, making them vulnerable to persistent texture attacks. However, optimized patterns can stand out from their surroundings, making an attack easier to notice and its carrier more likely to be inspected or removed. To address this limitation, our key insight is to search the structured image distribution of a diffusion model for natural-looking textures while using closed-loop feedback to identify task-disruptive candidates. Based on this idea, we propose DMT-Attack, a gradient-free gray-box attack that generates textures for ordinary scene objects such as tablecloths. A frozen text-conditioned diffusion model defines the search space, and an eight-expert latent proposal network is initialized from evaluated teacher latents and refined using rollout feedback. We evaluate DMT-Attack against pi_0.5 and OpenVLA-OFT across four LIBERO suites. DMT-Attack achieves a macro clean-conditioned attack success rate (cASR) of 57.9%, compared with 44.7% for UPA-RFAS and 19.7% for random natural textures. It also attains higher scene similarity and a lower clean-calibrated detector flag rate (45% versus 97.5-100% for optimized patch baselines). These results demonstrate that natural-looking textures can disrupt VLA visual feedback and closed-loop decision-making, motivating defenses that jointly assess visual appearance and task behavior.

cs.AI

新论文 · v1arXiv:2611.25553

TwoPass: Draft-Guided Few-Step Video Prediction for Robotic Manipulation

Yijun Liu, Shijia Ge, Tongtong Feng, Xin Wang, Yuan Meng, Yuwei Liu, Zhi Wang, Wenwu Zhu

原文摘要

Action-conditioned video prediction lets robots anticipate the visual consequences of their actions for planning and policy evaluation. Existing predictors denoise each future from noise without using their own estimate of it, which limits the quality of the predicted future, in particular of the manipulated objects, and requires many network calls. We present TwoPass, a self-conditioned training-and-inference framework that keeps the standard flow-matching objective. Self-Generated Draft Training supplies a detached clean-video estimate from the shared network as conditioning at the same noisy state. A Refinement-Only Late Adapter applies a dedicated residual transformation when a draft is present. At inference, Draft-Reuse Iterative Refinement obtains both a solver update and the next draft from each velocity prediction, requiring as few as 4 function evaluations (NFE) per video chunk. Training samples discrete solver-query noise levels and updates the pretrained backbone. Extensive experiments on Bridge and Language Table show that TwoPass achieves state-of-the-art results with 4 NFE, far fewer than the 50-59 calls of existing predictors. Code: https://github.com/PlantPotatoOnMoon/TwoPass.

cs.CV

版本更新 · v2arXiv:2611.24259

Do Thinking VLMs Need to Keep Looking at Long Videos? Write It Down, Then Look Away

Pengyu Li, Zhitao Gao, Lingling Zhang, Heng Wang, Chengqian Ma, WenHao Wang, Wenxuan Huang, Chen Zheng 等 9 位作者

原文摘要

Vision-language models (VLMs) reason about long videos over hundreds to thousands of tokens. Every decoding step re-reads the key-value (KV) cache of the entire video. Must the model keep looking while it reasons? In LookAway, the model first writes timestamped notes on what the question needs and then puts the video away. Once the notes close, the inference engine hides the video's KV cache for the rest of the response. Across both hybrid and pure-attention VLMs, dependence on the video is concentrated early in the response. On Qwen3.6-27B, writing notes preserves accuracy better than free reasoning at the same withdrawal step. Replacing them with another video's notes harms accuracy, whereas keeping the video visible alongside the model's own notes adds no significant benefit. Without additional training, LookAway nearly matches full-video thinking accuracy at 0.12-0.13 times its modeled decode cost. To shorten this initial look, we fine-tune the model on its own correct LookAway responses. Reinforcement learning then rewards correct answers while penalizing KV reads. The trained model puts the video away after a median of 66-68 tokens. Across three held-out benchmarks, it writes more tokens than an identically trained full-video model yet reads only 27-35% as many KV positions at matched accuracy.

cs.CV

版本更新 · v2arXiv:2611.24267

Look Closer, Look Back, or Wait for Agentic Streaming Video Understanding

Pengyu Li, Zhitao Gao, Lingling Zhang, Wenxuan Huang, YiCheng Xiao, Chen Zheng, Jun Liu

原文摘要

Streaming video models must answer questions about a live video under strict memory and latency limits. Feeding only the latest few frames keeps present perception sharp, whereas adding past frames dilutes attention on the current scene and triggers hallucinations. Recent cascade methods retrieve past frames when uncertain, but they still fail when current frames are blurry or when decisive future clues have not appeared yet. We introduce ActiveForager, a streaming video agent that maintains a clean short working window and gathers extra evidence only on demand. For each query, the agent either answers directly, zooms into high-resolution recent frames, recalls past events from memory, or waits for upcoming frames. When recalling history, a shared-weight sub-agent verifies candidate clips in isolation and returns a compact evidence summary. A stream-decoupled attention mask then keeps historical evidence from interfering with present visual tokens. We train this policy using counterfactual information gain, which rewards a tool call only when the gathered evidence makes the true answer more likely and prevents tool collapse across all actions. Across Qwen2.5-VL-7B and Qwen3-VL-8B on OVO-Bench, ActiveForager improves real-time perception, historical recall, and proactive response at once, lifting overall accuracy on Qwen3-VL-8B from 47.6% to 65.9%. It also doubles zero-shot accuracy on the multi-hop Video-Holmes benchmark while adding minimal visual tokens per query.

cs.CV

新论文 · v1arXiv:2611.25168

What a Calibrated Probability Hides: Belief-Level Post-Training for Decision Models

Xinye Li, Zirui Song, Huaxing Liu, Xiang Wang, Shuai Li, Zixiang Xu, Yinjun Liu, Zhewen Zhang 等 10 位作者

原文摘要

A calibrated decision model can predict how often each answer is chosen. The same prediction can reflect disagreement among people who are confident in their choices or uncertainty among people who share a belief. Disagreement calls for identifying whose perspective applies, while uncertainty calls for gathering more evidence. Standard post-training scores a single prediction against one observed choice at a time, learning the pooled distribution without distinguishing these two sources of uncertainty. We introduce Belief-Grouped Policy Optimization (BGPO). The policy proposes a list of beliefs, estimates their prevalence, and predicts a choice distribution for each. During training, BGPO uses Bayes' rule to estimate how likely each belief was to generate an observed choice. These estimates weight the choice's contribution to each belief's log-score reward. To discover rare beliefs, the policy also proposes lists after seeing a few observed choices and retains them only if they predict held-out choices well. At inference, the policy uses only the question and combines the belief-specific distributions by prevalence to produce a calibrated aggregate prediction. On benchmarks covering opinions, conversation safety, emotions elicited by artwork, and photo aesthetics, BGPO achieves the lowest conditional, worst-group, and overall Jensen-Shannon divergence across model initializations. It improves conditional divergence on both text and image tasks. On a synthetic image task with known beliefs, BGPO recovers the underlying groups more accurately than baselines trained with belief labels.

cs.CV

新论文 · v1arXiv:2611.24864

NereidVLA:学习理解与导航水下世界

NereidVLA: Learning to Understand and Navigate the Underwater World

Chenjie Zhao, Zhiliang Zhu, Quan Chen, Yuquan Xue, Xingxue Shi, Puyue Wang, Zhiqiang Zhao, Kangye Zheng 等 11 位作者

AI 速读 基于摘要

作者提出 NereidVLA,先以约两百万海洋样本进行海洋空间预训练(Sea2Know),再用仿真水下动力学下的专家轨迹适配,使视觉-语言-动作模型能根据语言指令预测离散运动 token 并解码为局部 3D 轨迹,用于闭环水下导航。

cs.CV · cs.RO

查看原文摘要

Underwater navigation requires translating visual spatial cues into motion, yet marine vision-language datasets rarely include paired robot trajectories. We present NereidVLA, a vision-language-action model that transfers marine spatial knowledge to language-conditioned 3D navigation. We first train Sea2Know on approximately two million marine examples to learn target grounding, spatial relations, and scene structure without paired robot-motion supervision. We then adapt it using expert trajectories executed under simulated underwater vehicle dynamics. Given RGB histories and a language instruction, NereidVLA predicts discrete motion tokens that decode into local 3D trajectories for closed-loop tracking. Evaluation combines an independent six-dimensional underwater navigation understanding benchmark with offline prediction and closed-loop navigation. Under matched navigation training and execution settings, marine spatial pretraining increases simulation success from 72.4% to 84.4%. Without fine-tuning on the real test environment, NereidVLA achieves 78.75% success in laboratory-pool robot trials, compared with 47.50% for InternVLA-N1. These results show that marine spatial supervision can support downstream motion learning and physical underwater navigation.

新论文 · v1arXiv:2611.24220

Erased but Recoverable: On the Fragility of Concept Erasure Under Fine-Tuning

Jun Li, Lizhi Xiong, Ziqiang Li, Yu Liu, Zenglin Shi

原文摘要

Existing concept erasure methods for text-to-image diffusion models primarily focus on suppressing target concepts while preserving non-target generation quality. However, in practical applications, open-source diffusion models are often further adapted through DreamBooth, LoRA, or Custom Diffusion. Such adaptations can reshape internal representations, thereby weakening or even reversing the erasure effects achieved by existing methods. We identify erasure persistence as a key yet underexplored dimension of concept erasure and analyze how downstream fine-tuning reshapes internal representations. Our analysis reveals that fine-tuning of the original model does not uniformly modify the model, but preserves certain high-contribution structures, particularly attention heads with stable output subspaces and response magnitudes. We also observe that, after downstream fine-tuning of concept-erased models, these heads' output representations more readily realign with their pretrained counterparts. Motivated by this observation, we propose Stable-Head-Aware Concept Erasure (SHACE), which improves erasure persistence by preserving stable attention structures during concept erasure. SHACE identifies stability-critical attention heads through Attention-Head Stability Profiling and keeps their corresponding parameter blocks unchanged during erasure via Stable-Head Gradient Preservation. Extensive experiments show that SHACE improves resistance to concept recovery across diverse fine-tuning settings while maintaining strong suppression and non-target generation quality.

cs.CR

版本更新 · v2arXiv:2611.22245

Multi-Agent Memory Inconsistency: Where Factual Conflicts Disappear in LLM Agents

Hejia Zhang, Zhongming Yu, Yimin Ding, Wei-Ting Chen, Hengjia Yu, Wentao Ni, Yijun Liu, Cong Xu 等 12 位作者

原文摘要

Multi-agent delegation lets LLM systems distribute evidence collection across workers with separate contexts. This creates a consistency challenge: facts retrieved by workers may be missing from the evidence available to the lead for a joint conclusion. We introduce a benchmark and evidence-flow profiler for studying multi-agent memory inconsistency through controlled factual conflicts in Harvey LAB document-review tasks. Pooled across three models with comparable generation spending, requiring delegation increases retrieval of both conflicting claims from 50.2% to 68.7% relative to single-agent execution, yet conflict detection falls from 16.2% to 13.2%. Within this setting, workers report both claims for only 35.2% of retrieved pairs, and pooled matched comparisons show higher detection than single-agent execution when one worker reads both claims but lower detection when the claims are split across workers. Worker-to-lead handoffs are a central target for improving conflict detection in multi-agent LLMs.

cs.AI · cs.MA

新论文 · v1arXiv:2611.23930

A Hierarchical Onboard Vision-Language-Navigation Framework for Aerial Robots in Complex Environments

Yuze Wu, Mo Zhu, Xingxing Li, Yuheng Du, Zhaoqi Wang, Yuxin Fan, Wenjun Li, Kenghou Hoi 等 11 位作者

原文摘要

Unmanned Aerial Vehicles (UAVs) are expected to achieve semantic reasoning, intelligent decision-making, and long-horizon navigation under strict onboard constraints. While recent foundation-model-driven approaches have shown promising progress toward this goal, existing methods remain limited by challenges in 3D data availability, trajectory feasibility, long-horizon reasoning, and computational efficiency in complex environments. To bridge these gaps, we propose VLN-AR, a hierarchical vision-language-navigation (VLN) framework for aerial robots. VLN-AR decouples multimodal reasoning and low-level control, enabling structured prediction of 3D pose commands and navigation intentions that are translated into dynamically feasible actions. To mitigate data scarcity and domain gaps, we introduce a fully automated 3D Gaussian Splatting (3DGS)-based pipeline to generate scalable photorealistic UAV navigation data. A progressive three-stage training strategy further enhances grounding, scene reasoning, and complex navigation. In addition, a lightweight cognitive agent improves long-horizon reasoning via memory-aware task decomposition. Finally, we design onboard optimization techniques, including token sparsification and model lightweighting, achieving 7.1x-9.4x speedup while maintaining high navigation performance on resource-limited UAV platforms. Experiments show great improvements over existing methods in both simulation and real-world UAV navigation. Project website: https://wyz000.github.io/VLN-AR/

cs.RO · cs.AI

版本更新 · v2arXiv:2610.00262

Risk-Calibrated Intent Routing for Conversational AI via Signed Lexical Evidence

Yezhou Cheng, Zehua Yang, Bojun Lin

原文摘要

Conversational AI has become widely used as advances in language models let people find information, get writing help, and ask for practical guidance in their own words. When an assistant handles specific tasks, it must infer what the user wants and decide whether to act on that prediction. A sentence classifier's confidence can miss clues in the wording. We combine the gap between the classifier's two highest scores with signed lexical evidence, a score based on words and short character patterns. The score is positive when those patterns favor the predicted intent and negative when they favor another. We use separate calibration data and a statistical upper bound on the error rate to set the minimum confidence required to accept a request. Across ten runs per dataset, our method reduces the area under the risk-coverage curve, which summarizes error as more requests are accepted, by 15.8%, 15.1%, and 11.8% on BANKING77, CLINC150, and HWU64, respectively, compared with a confidence model that uses only sentence classifier scores. At a 5% error target, the share of accepted requests rises by 1.83 percentage points on BANKING77 and 5.14 points on HWU64. CLINC150 already has near-full coverage at that target. At a 2% target, calibration finds a rule that accepts some requests in all 30 runs. Our contribution is a two-feature confidence model that improves routing by measuring how strongly the wording supports or challenges the original intent prediction.

cs.CL · cs.AI

版本更新 · v2arXiv:2611.13662

NIF-VIO: Visual--Inertial Odometry Across Inertial and Non-Inertial Reference Frames

Zichao Shu, Lijun Li, Zhuochen Li, Zetao Chen, Jianyu Wang

原文摘要

Conventional visual--inertial odometry (VIO) assumes that observations from cameras and an inertial measurement unit (IMU) are jointly consistent in an inertial reference frame. However, this assumption breaks down under non-inertial conditions. Inside accelerating or rotating carriers, visual observations of the interior reflect carrier-relative motion, while IMU measurements contain additional carrier-induced components, which can compromise conventional fusion. We therefore present NIF-VIO for localization relative to the local environment across inertial and non-inertial reference frames. It adapts visual--inertial fusion based on measurement consistency. As inertial conditions change, early constraint isolation is combined with local reoptimization to limit error propagation before inconsistency confirmation and repair the affected local geometry. Under non-inertial conditions, a compact model with temporally shared parameters is introduced to account for carrier motion while preserving inertial information that complements visual constraints. Experiments on public benchmarks and a self-collected dataset demonstrate that NIF-VIO achieves competitive performance under inertial conditions and improved localization accuracy and availability under non-inertial conditions.

cs.RO

新论文 · v1arXiv:2611.23719

Improving While Deployed: Learning Proactive End-to-End Driving from Imperfect Human Takeover Data

Yangang Ren, Di Zeng, Jun Li, Zirui Li, Chengjiang Zhang, Xiaoqing Ye, Ji Tao, Chen Lv

原文摘要

Once deployed on vehicles, end-to-end driving models continuously encounter challenging scenarios, generating valuable human takeover data that capture both model outputs and human behaviors. However, human interventions are often imperfect, as drivers may intervene too late or unnecessarily. Consequently, direct imitation can reinforce late responses or discourage otherwise appropriate maneuvers. To address this challenge, we propose an offline post-training framework that transforms imperfect takeover data into proactive supervision. We first construct diverse trajectory candidates by augmenting the plans of the deployed model and shifting recorded human corrective maneuvers earlier in time. We then train a scene-conditioned trajectory scorer to rank these candidates using expert demonstrations and unpaired preference signals from takeover logs. Using this scorer, we selectively replace pre-takeover training targets with higher-scoring candidates while retaining the original targets in normal driving scenarios. Finally, we distill the resulting targets into the driving model without changing its inference architecture. In closed-loop evaluations using real-world scenarios reconstructed with 4D Gaussian splatting, our method achieves the highest overall success rate, outperforming the deployed model, direct imitation learning and trajectory post-optimization. These results highlight a practical path toward efficient updates of deployed driving models using existing fleet logs.

cs.RO

版本更新 · v2arXiv:2610.32433

Tempo-Zero: Recursive Self-Play in Executable Worlds for Video Reasoning

Jingmin Zhu, Yanyan Zou, Xiuyu Li, James Bailey, Dinh Phung, Tao Yu, Yunyun Hou, Jun Liu 等 13 位作者

原文摘要

Video reasoning requires understanding how events unfold over time, yet videos with accurate temporal annotations are expensive to collect. Self-play offers an alternative by letting a model generate and solve its own tasks without manual annotation. However, existing video self-play methods train on fixed sets of recorded videos, so they cannot alter the underlying event dynamics and rely on noisy pseudo-labels derived from the model's own outputs, such as majority voting, that risk reinforcing shared model errors. To overcome these limitations, we introduce Tempo-Zero, a self-improvement framework that closes a loop of executable environments, execution-verified self-play, and environment evolution, where executing a program renders a video with the ground-truth option and interval of every question. A single policy plays both Questioner and Solver, proposing questions and answering them with timestamped evidence lines, and is trained on execution-grounded rewards instead of model consensus. The Solver's performance steers a curriculum over tasks and difficulty levels, and the policy selects rule changes that enrich the event dynamics of later rounds. Despite being trained solely on simple programs, without supervised fine-tuning or external video data which prior methods typically require, Tempo-Zero outperforms the base models and recent self-play baselines on average across twelve real-world video benchmarks at different model scales, with particularly strong gains on temporal grounding tasks. Project page: https://alchemist-jo.github.io/tempo-zero/

cs.CV

版本更新 · v2arXiv:2609.29140

Cost-Efficient LLM Evaluation: A Framework for Narrow Confidence Intervals Under Fixed Budgets

Yezhou Cheng, Runjia Du, Zeming Liu, Qibai Chen, Hang Lyu, Yilan Wei, Yankai Zeng, Bojun Lin

原文摘要

Large language models (LLMs) play an increasingly important role in both academic research and industry. Reliable evaluation helps researchers understand model capabilities and assess progress during model development. However, evaluation can be costly, and researchers often need to assess performance within a fixed budget. Because model outputs can vary across runs, this budget must be allocated between testing different tasks and repeating selected tasks. A confidence interval quantifies the uncertainty in the resulting performance estimate. Obtaining narrower intervals at the same confidence level and evaluation cost provides a more precise assessment of model performance, helping researchers draw clearer conclusions from limited observations. We introduce a framework that allocates a fixed evaluation budget across tasks and repeated runs, with or without full task coverage. The framework then uses disagreements between runs on the same task to bound estimation error, allowing narrower confidence intervals at the same confidence level when the sampled repeats show little variability. When repeated runs on each task yield the same outcome, we prove that our framework achieves the fastest theoretically possible rates for narrowing confidence intervals in the worst case as the number of tasks and the budget for repeats grow. We establish matching lower bounds both when every task must be covered and when tasks may be skipped, even for evaluators who use earlier results to decide which run to carry out next. On LiveCodeBench, our framework yielded narrower confidence intervals on average than uniform random sampling in 15 of 16 model configurations, with a median width reduction of 30.6% at the same confidence level and the same budget of 1,100 outputs per configuration.

cs.AI · cs.IT

新论文 · v1arXiv:2611.23005

CARE: Conversational Affective Response Retrieval for Internet Memes

Jiajun Li, Leihan Zhang, Teng Zhang, Le Zhang, Qiang Yan

原文摘要

Conversational meme retrieval supports response selection in online communication, where relevance depends on both contextual meaning and affective expression. We study this task using two language-specific collections from Douban and Imgflip, containing 98,506 meme records across 5,583 discussion topics. Thread membership supplies weak relevance labels, while likes provide implicit ranking feedback. Automatically extracted sentiment, facial-expression indicators, and cross-modal consistency features support an exploratory analysis of engagement. We propose CARE, a Conversational Affective Response Retrieval framework that combines an emotion-enhanced CLIP retriever with a Jina-based reranker. The retriever incorporates free-form emotion descriptors through gated fusion and learns from multiple positive examples per query. The reranker combines query--image interactions with emotion representations and is trained using engagement-based ranking supervision. Under the reported topic-held-out protocol, CARE achieves Hit@10 of 24.03\% on Douban and 49.33\% on Imgflip, compared with 23.16\% and 41.07\%, respectively, for fine-tuned CLIP. These results support the utility of the complete two-stage system for recovering thread-associated memes. They do not independently establish perceived social resonance or isolate the contribution of affect from reranker training. The study provides a basis for further evaluation of affect-aware retrieval with human relevance judgments and controlled comparisons.

cs.IR

版本更新 · v3arXiv:2507.02844

Visual Contextual Attack: Jailbreaking MLLMs with Image-Driven Context Injection

Ziqi Miao, Yi Ding, Chen Qian, Yongyao Wang, Zaoying Huang, Yuan Xiong, Ken Deng, Lijun Li 等 9 位作者

原文摘要

With the emergence of strong vision language capabilities, multimodal large language models (MLLMs) have demonstrated tremendous potential for real-world applications. However, the security vulnerabilities exhibited by the visual modality pose significant challenges to deploying such models in open-world environments. Recent studies have successfully induced harmful responses from target MLLMs by encoding harmful textual semantics directly into visual inputs. However, in these approaches, the visual modality primarily serves as a trigger for unsafe behavior, often exhibiting semantic ambiguity and lacking grounding in realistic scenarios. In this work, we define a novel setting: vision-centric jailbreak, where visual information serves as a necessary component in constructing a complete and realistic jailbreak context. Building on this setting, we propose the VisCo (Visual Contextual) Attack. VisCo fabricates contextual dialogue using four distinct vision-focused strategies, dynamically generating auxiliary images when necessary to construct a vision-centric jailbreak scenario. To maximize attack effectiveness, it incorporates automatic toxicity obfuscation and semantic refinement to produce a final attack prompt that reliably triggers harmful responses from the target black-box MLLMs. Specifically, VisCo achieves a toxicity score of 4.78 and an Attack Success Rate (ASR) of 85% on MM-SafetyBench against GPT-4o, significantly outperforming the baseline, which achieves a toxicity score of 2.48 and an ASR of 22.2%. Code: https://github.com/Dtc7w3PQ/Visco-Attack.

cs.CV · cs.CL · cs.CR

版本更新 · v3arXiv:2507.05248

Response Attack: Exploiting Contextual Priming to Jailbreak Large Language Models

Ziqi Miao, Lijun Li, Yuan Xiong, Zhenhua Liu, Pengyu Zhu, Chen Qian, Haonan Jia, Zihang Xu 等 10 位作者

原文摘要

Contextual priming, where earlier stimuli covertly bias later judgments, offers an unexplored attack surface for large language models (LLMs). We uncover a contextual priming vulnerability in which the previous response in the dialogue can steer its subsequent behavior toward policy-violating content. While existing jailbreak attacks largely rely on single-turn or multi-turn prompt manipulations, or inject static in-context examples, these methods suffer from limited effectiveness, inefficiency, or semantic drift. We introduce Response Attack (RA), a novel framework that strategically leverages intermediate, mildly harmful responses as contextual primers within a dialogue. By reformulating harmful queries and injecting these intermediate responses before issuing a targeted trigger prompt, RA exploits a previously overlooked vulnerability in LLMs. Extensive experiments across eight state-of-the-art LLMs show that RA consistently achieves significantly higher attack success rates than nine leading jailbreak baselines. Our results demonstrate that the success of RA is directly attributable to the strategic use of intermediate responses, which induce models to generate more explicit and relevant harmful content while maintaining stealth, efficiency, and fidelity to the original query. The code and data are available at https://github.com/Dtc7w3PQ/Response-Attack.

cs.CL

版本更新 · v3arXiv:2603.28618

Seeing with You: Perception-Reasoning Coevolution for Multimodal Reasoning

Ziqi Miao, Haonan Jia, Lijun Li, Chen Qian, Yuan Xiong, Wenting Yan, Yongyao Wang, Zaoying Huang 等 10 位作者

原文摘要

Reinforcement learning with verifiable rewards (RLVR) has substantially enhanced the reasoning capabilities of multimodal large language models (MLLMs). However, existing RLVR approaches typically rely on outcome-driven optimization that updates both perception and reasoning using a shared reward based solely on the final answer. This shared reward blurs credit assignment, frequently improving reasoning patterns while failing to reliably enhance the accuracy of upstream visual evidence extraction. To address this perception bottleneck, we introduce PRCO (Perception-Reasoning Coevolution), a dual-role RLVR framework with a shared policy. PRCO consists of two cooperative roles: an Observer that generates an evidence caption tailored to the question and a Solver that predicts the final answer based on this caption. Crucially, PRCO employs role-specific reward signals: the Solver is optimized using verifiable outcome rewards on the final answer, while the Observer receives a utility reward derived from the Solver's downstream success. Extensive experiments across eight challenging multimodal reasoning benchmarks demonstrate that PRCO yields consistent improvements across model scales by over 7 points on average accuracy compared to the base model, outperforming prior open-source RL-tuned baselines.

cs.AI

新论文 · v1arXiv:2611.22686

CascadeJEPA: Enhancing 3D Geometric Representations via Spatially Coupled Cascaded Prediction

Kailai Zhou, Ziqi Chen, Xiaoning Xu, Yuefeng Wu, Zixuan Fu, Zijun Lin, Zhihao Li, Youwei Pang 等 11 位作者

原文摘要

Joint-Embedding Predictive Architectures (JEPAs) have emerged as a promising paradigm for predictive visual representation learning. However, our analysis of I-JEPA reveals a growing discrepancy between semantic and geometric learning: improved semantic recognition during pretraining is accompanied by degraded 3D geometric representations. We hypothesize that this partly stems from predicting highly abstract latent representations with incomplete spatial context, favoring semantic regularities over fine-grained geometric information. To address this issue, we introduce CascadeJEPA, a spatially coupled cascaded prediction framework that promotes richer patch-level interactions through full-grid context encoding and cross-attention between prediction stages, while largely preserving the original I-JEPA architecture. Extensive experiments demonstrate substantial improvements over I-JEPA across multiple 3D geometric tasks while maintaining strong semantic recognition performance. Furthermore, its advantages extend to future feature forecasting and world model planning, highlighting the potential of enhanced geometric representations to support predictive world modeling.

cs.CV

版本更新 · v2arXiv:2610.40339

Wizard: Manipulate Any World in Real Time via Agentic Control

Yuzhe Wu, Zijun Lin, Cheston Tan, Bihan Wen, Yeying Jin

原文摘要

Recent advances in video world models have enabled increasingly realistic and interactive virtual environments, opening new possibilities for player-customizable gameplay. Agentic frameworks further advance this direction by decoupling explicit game logic and controllable world representations from visual generation. However, existing approaches support only simple game settings and predefined motion patterns, generalize poorly to unseen environments and characters, and offer limited real-time interactivity. In this paper, we introduce Wizard, an agentic game world model that enables real-time, precise control over character motions while generalizing zero-shot across novel scenes, characters, and actions. To achieve this, we construct a diverse dataset spanning 805 scenes and 178 characters, incorporating skeleton and geometry information as structured proxies to condition video generation. Our agentic framework explicitly tracks and updates game states to support character-aware adaptive masking and world-anchored spatial conditioning, improving character identity preservation and spatial consistency. Experiments on WBench show that our method achieves higher gameplay fidelity with comparable video quality and spatiotemporal consistency, while using fewer parameters and less training data than state-of-the-art world models. These advantages are also retained in our two-step real-time variant. Overall, Wizard combines agentic control with structured proxy-based generation to enable player-driven game creation with real-time, generalizable, and mechanically consistent world simulation.

cs.CV

新论文 · v1arXiv:2611.21806

A Probabilistic Lateral Motion Prediction Model with Improved Response Fidelity for Vehicle-in-the-Loop Simulation

Xiangyang Ye, Chaosheng Huang, Jinwen Yang, Zhengxian Chen, Jun Li

原文摘要

Vehicle-in-the-loop (VIL) simulation requires a lateral dynamics model that accurately reproduces vehicle responses and reliably generates horizontal lateral-position and heading commands through recursive integration. This paper presents a probabilistic lateral motion prediction model with improved response fidelity. A two-degree-of-freedom vehicle model provides the physical prior, while a recurrent neural network learns correlated residual corrections to lateral acceleration and yaw rate. To maintain prediction fidelity during recursive motion generation, multistep rollout training feeds corrected responses into subsequent states and histories and directly penalizes accumulated lateral-velocity error. A conditional Gaussian characterizes response uncertainty, which is propagated to lateral-response, position, and heading predictions through augmented-state linearization. Evaluation results demonstrate improved lateral-response and integrated-motion prediction. Compared with the Deep Dynamics lateral adaptation, the proposed model reduces mean lateral-position root-mean-square error (RMSE) from 0.596 m to 0.483 m, an improvement of 19.0%, while reducing mean lateral-velocity RMSE from 0.0469 m/s to 0.0441 m/s. These results demonstrate the effectiveness of combining physical priors, probabilistic residual learning, and multistep rollout training for response-faithful lateral motion prediction.

cs.RO

版本更新 · v3arXiv:2603.05963

Skeleton-to-Image Encoding: Enabling Skeleton Representation Learning via Vision-Pretrained Models

Siyuan Yang, Jun Liu, Hao Cheng, Chong Wang, Shijian Lu, Hedvig Kjellstrom, Weisi Lin, Alex C. Kot

原文摘要

Recent advances in large-scale pretrained vision models have demonstrated impressive capabilities across a wide range of downstream tasks, including cross-modal and multi-modal scenarios. However, their direct application to 3D human skeleton data remains challenging due to fundamental differences in data format. Moreover, the scarcity of large-scale skeleton datasets and the need to incorporate skeleton data into multi-modal action recognition without introducing additional model branches present significant research opportunities. To address these challenges, we introduce Skeleton-to-Image Encoding (S2I), a novel representation that transforms skeleton sequences into image-like data by partitioning and arranging joints based on body-part semantics and resizing to standardized image dimensions. This encoding enables, for the first time, the use of powerful vision-pretrained models for self-supervised skeleton representation learning, effectively transferring rich visual-domain knowledge to skeleton analysis. While existing skeleton methods often design models tailored to specific, homogeneous skeleton formats, they overlook the structural heterogeneity that naturally arises from diverse data sources. In contrast, our S2I representation offers a unified image-like format that naturally accommodates heterogeneous skeleton data. Extensive experiments on NTU-60, NTU-120, and PKU-MMD demonstrate the effectiveness and generalizability of our method for self-supervised skeleton representation learning, including under challenging cross-format evaluation settings.

cs.CV · cs.AI

版本更新 · v2arXiv:2603.27850

Automated Code Efficiency Optimization via Agent Skill Mining

Zimu Wang, Yuling Shi, Mengfan Li, Zijun Liu, Jie M. Zhang, Chengcheng Wan, Xiaodong Gu

原文摘要

Code efficiency is a fundamental aspect of software quality, yet how to use large language models (LLMs) to optimize correct programs remains challenging. Prior approaches rely on one-shot rewriting, retrieved exemplars or instructions, fine-tuning, or candidate search. In this paper, we present EffiSkill, a two-stage framework that models recurring slow-to-fast transformations as Operator Skills: persistent cards that record a transformation's intent, applicability, procedure, expected effect, and risks. Stage I distills frozen, language-specific skill libraries from slow/fast program pairs. Stage II diagnoses a new program, retrieves relevant skills, composes alternative plans, and generates up to eight candidates. These steps finish without candidate-execution feedback; public tests then filter and rank candidates, and held-out private tests determine reported correctness and speedup. Across 623 EffiBench-X tasks per language, two backbones, and separate Python and C++ evaluations, EffiSkill achieves the highest observed OPT@1 and OPT@8 among the compared methods in all four settings. Its OPT@8 margins over the strongest baselines range from 4.98 to 12.52 percentage points. These results support broader successful optimization coverage for the complete skill-guided pipeline under a shared output budget.

cs.SE · cs.CL

新论文 · v1arXiv:2611.19771

JEPA4R: Embodied JEPA World Models with 4D Point Planning

Jiarui Dong, Zhiheng Liang, Yanjun Li, Zeyu Zhang, Heejeong Nam, Haoyu Chen, Hao Tang

原文摘要

Joint-embedding predictive architecture (JEPA) world models predict future embeddings instead of pixels and plan in latent space. Two challenges remain: 2D images leave the metric geometry that manipulation acts on implicit and entangled with appearance, and what these models imagine stays hidden, as they are mostly judged in simulation by short-horizon latent planning. We present JEPA4R, a JEPA world model that observes single-depth-camera point clouds, predicts and plans in their latent space, and acts through a policy head on its frozen latents; we call this 4D point planning. A decoder on the frozen world model shows what it imagines as 4D point clouds, with a mean point error within 1.1 cm over 200 steps. Under the policy-learning protocols of OGBench and CortexBench, JEPA4R leads pixel JEPA models on OGBench Cube-Single, Cube-Double, Cube-Triple and Scene, and ranks first on MetaWorld and Adroit Relocate with 98.4% and 88.0%, above the runner-up, R3M, at 96.0% and 65.3%. Trained only in a digital twin, JEPA4R and its policy heads run zero-shot on the real arm with both a gripper and a dexterous hand.

cs.RO · cs.CV

版本更新 · v2arXiv:2611.17048

Beyond Answering: Learning to Speak First and Act in Streaming Video

Shuhan Li, Huankang Guan, Xintong Guo, Yang Bo, Yuyang He, Dunyuan Xu, Jingyan Chen, Fang Liu 等 18 位作者

原文摘要

Most AI assistants respond only when prompted, remaining silent even when events in the video stream offer meaningful opportunities for conversation. To move beyond purely user-driven interaction, we present AURA 2.0, a streaming video large language model that introduces three core upgrades to AURA. The key advance is Model-Initiated Conversation (MIC), which enables the model to identify topics worth discussing in the evolving visual stream and initiate conversations at appropriate moments without a user query. Stream-Aware Tool Calling (SATC) lets the model act on what it sees by invoking external tools using the current visual and conversational context and incorporating returned information into the ongoing interaction. Interactive Visual Grounding (IVG) supports visual grounding in both user inputs and model outputs. The model can identify objects marked by users on video frames and produce frame-aligned bounding boxes to link its responses to specific objects and regions. To develop these capabilities, we construct a dedicated streaming interaction dataset and adopt a three-stage training scheme to integrate the new behaviors with AURA's existing streaming interaction abilities. We further incorporate Long-Horizon Interaction Memory (LHIM) to maintain contextual continuity over extended video streams. Evaluations on public streaming benchmarks show that AURA 2.0 achieves leading performance among open-source streaming models, while targeted evaluations further validate its new interaction capabilities.

cs.CV · cs.AI

版本更新 · v2arXiv:2610.41180

Quantum query complexity of Hurwitz stability testing

Maxwell Fitzsimmons, Mohamed Serry, Meenu Kumari, Jun Liu

原文摘要

Determining whether every trajectory of the linear system $\dot{x}=Ax$ converges to zero (or, equivalently, whether $A$ is Hurwitz stable) is a fundamental problem in control theory. It arises, for example, in verifying the stability of feedback controllers and state observers, as well as of linearized dynamical models. We study the quantum query complexity of deciding whether a matrix $A\in\mathbb{C}^{n\times n}$ is Hurwitz stable, given an exact normalized block-encoding of $A$. The query complexity is characterized in terms of the system dimension $n$ and a supplied lower bound $0<ν\leq ω_A$ on the imaginary-axis distance $ω_A:=\min_{t\in\mathbb{R}}σ_{\min}(A-\mathrm{i}tI)$, where $σ_{\min}(A)$ is the smallest singular value of $A$. We prove a query-complexity lower bound of $Ω(\sqrt{n}/ν)$, which holds even when $A$ is restricted to be real and diagonal. For general matrices with nonzero $ω_A$, we encode an approximation of the right-half-plane spectral projector $\operatorname{sign}^+(A):=(I+\operatorname{sign}(A))/2$ as a block of a matrix inverse, where $A$ is Hurwitz if and only if the projector is zero. To test this, we produce a quantum algorithm which determines if a $(β,b,1/(4\sqrt{n}))$-block-encoded projector matrix (like $\operatorname{sign}^+(A)$) is zero or not; for constant success probability, it has complexity $Θ(β\sqrt{n})$. When the zero test is applied to specific approximations of $\operatorname{sign}^+(A)$, this yields an upper bound of $O(\sqrt{n}ν^{-2}\log(\sqrt{n}/ν))$ queries in the general case, and $O(\sqrt{n}ν^{-1}\log n)$ when $A$ is Hermitian or normal. The latter upper bound matches the lower bound up to a logarithmic factor. Thus the dimension dependence of quantum Hurwitz testing is nearly optimal, while the remaining factor in the non-normal case isolates an open conditioning gap.

quant-ph · cs.CC

新论文 · v1arXiv:2611.18183

When Should a Virtual Patient Ask? RISP-Guided Proactive Information Seeking in Multi-Agent Patient Simulation

Jingyao Cai, Shuaijun Liu, Abdul Rehman, Xiaowan Li, Qintian, Thomas Dolby, Sue M Green, Xiaosong Yang

原文摘要

Virtual patient (VP) systems support nursing communication training by providing simulated encounters in which learners practise explaining care and responding to patient concerns. Although large language models (LLMs) support flexible dialogue, current approaches give less explicit attention to how patients' information needs develop and guide the content and timing of their questions. We draw on the Risk Information Seeking and Processing (RISP) model, which relates information seeking to perceived risk and the gap between perceived knowledge and desired understanding. Guided by this perspective, we propose the RISP Multi-Agent System (RISP-MAS), a VP system for patient-initiated information seeking. The system links ongoing assessment of the patient's perceptions and information needs to coordinated decisions about whether to seek information, when to ask and which uncertainty to address. Evaluation of nursing student-VP interactions suggested that RISP-MAS exhibited plausible patient information-seeking behaviour, as indicated by expert assessments, while student feedback reflected positive interaction experiences. This work extends patient simulation by connecting a psychological account of information needs to explicit decisions about question initiation. The approach offers nursing students opportunities to practise recognising unresolved patient concerns and tailoring explanations to patients' information needs.

cs.MA

版本更新 · v2arXiv:2607.07824

CPM-MultiAgent: A CPM-Grounded Appraisal Framework for Simulating Dynamic Patient Emotional Evolution in Healthcare Dialogue

Jingyao Cai, Shuaijun Liu, Xiaowan Li, Yutong Guo, Abdul Rehman, Qintian, Thomas Dolby, Sue M Green 等 9 位作者

原文摘要

Large Language Models (LLMs) enable patient simulation for healthcare communication training, where learners practice responding to clinical information, concerns, and emotional reactions. However, existing patient simulation and affective dialogue systems often represent patient emotions as static attributes or focus on recognizing emotions and generating empathetic responses. They provide limited support for modeling how a simulated patient's emotional state evolves in response to a healthcare professional's communication. Changes in patient emotion constitute interactional feedback that learners must recognize and respond to across dialogue turns. We draw on the Component Process Model (CPM), a psychological theory that explains emotion as a dynamic process shaped by event appraisal. We propose CPM-MultiAgent, a CPM-grounded multi-agent framework for dynamic patient emotion simulation in healthcare communication training. The framework uses an emotion taxonomy informed by medical literature and expert input, and updates emotional states through healthcare-related trigger identification, structured CPM appraisal, and emotion-state integration, accounting for the patient's profile, clinical context, and prior emotional state. We evaluate the framework through student-led conversations and controlled trigger cases, combining nursing-expert assessments, LLM-judge comparisons, component ablations, backbone checks, and student feedback.

cs.MA · cs.AI

版本更新 · v2arXiv:2611.12762

Sage-Taiyi: A Metacognitive Architecture for Scientific Discovery

Weida Wang, Benteng Chen, Yigeng Jiang, Shuchen Pu, Weiping Shen, Zichao Liu, Yulin Su, Zitong Wang 等 90 位作者

原文摘要

We introduce Sage-Taiyi, a metacognition-inspired architecture that uses accumulated scientific understanding to guide what to investigate next and how to investigate it. Through explicit monitoring and control, the system revises its understanding of evidence, assumptions, and unresolved questions, then uses these assessments to adapt research targets and select reusable reasoning methods. On Taiyi-Bench, an AI co-scientist ideation benchmark comprising 265 tasks from 50 papers across ten disciplines, Sage-Taiyi ranks first on all eight metrics of historical rediscovery and open-ended ideation, achieving an Overall score of 54.0 compared with 47.9 for the strongest baseline, AI Scientist-v2. It also performs well on two external scientific ideation benchmarks. Five expert-guided collaborations further connect its proposals to experimental and computational validation across chemistry, materials science, and engineering. In chemistry, reaction-condition optimization raises the yield of cobalt-catalyzed carbonylation from trace amounts to 84% and photocatalytic allylic amination from 13% to 69%. These results support metacognitive regulation as an approach to scientific ideation and its progression toward practical validation.

cs.AI