cs.AI（2026-10-09）

📊 共 36 篇论文 | 🔗 6 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (18 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (15 🔗3) 支柱一：机器人控制 (Robot Control) (2) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (18 篇)

#题目一句话要点标签🔗⭐
1 UkrMeme: Multimodal Explainable Persuasion Detection in Ukrainian Memes 提出UkrMeme以解决乌克兰网络迷因中的说服技术检测问题 multimodal
2 Retrieval-Augmented Multimodal Language Model with Heterogeneous Evidence Integration for Clinical Prognosis 提出PRISM以解决临床预后中的多模态证据整合问题 multimodal
3 Agentic Large Language Models 提出代理型大型语言模型以实现自主行动与交互 large language model
4 Locate, Judge, Read: Role-Specialized Tool Agents over a Knowledge-Point Index for Open-Domain Multimodal Document QA 提出KnowForge以解决开放领域多模态文档问答问题 multimodal ✅
5 Learning More by Training Less: Uncertainty-Guided Progressive Contraction for Efficient Reasoning Post-Training 提出不确定性引导的渐进收缩框架以提高推理效率 large language model chain-of-thought
6 A Multi-Agent LLM System for Automated Tuning of Specialized Neural Network Models 提出多智能体LLM系统以自动调优专用神经网络模型 large language model
7 Same Query, Different Answer: Run-to-Run Variance in Semantic Query Processing 提出新方法以解决语义查询处理中的执行变异问题 large language model
8 SWE-DeceptionBench: Benchmarking Deceptive Behaviors of LLMs Across the Software Development Lifecycle 提出SWE-DeceptionBench以评估LLM在软件开发生命周期中的欺骗行为 large language model
9 AudioTETRIS: Token Elimination via Transitional Importance Signals for Efficient Audio LLMs 提出AudioTETRIS以解决音频语言模型的高计算成本问题 multimodal
10 Web$hell: A Unified Interface for Agentic Web Access 提出Web$hell以解决现有网页访问模型的局限性 large language model
11 Agent4Spec: An LLM-based Agentic System to Optimize API Specifications through Automated Validation and Market Analysis 提出Agent4Spec以优化API规范的自动验证与市场分析 large language model
12 VIGOR: Verifiable Industrial Geometry Optimization and Repair in Scene Programs 提出VIGOR以解决工业场景程序的可执行性与工程可行性差距问题 large language model
13 When Faster Decisions Do Not Make Better Agents: An Empirical Study of Jev in Agent Workflows 研究Jev在代理工作流中的效率转移问题 large language model
14 Attention Sees, MLP Refuses: Tracing the Attention-to-MLP Safety Pathway in LLMs 提出安全路径追踪方法以解决LLMs的拒绝问题 large language model
15 Travel Back To Move Forward: Retrospective Exploration and Evolving for Long Horizon Agents 提出ReTRAIL框架以解决长时间任务中的探索与回溯问题 large language model ✅
16 Work2Skill: Capturing Human GUI Experience for Reusable Agent Skill Induction 提出Work2Skill框架以解决人机交互技能重用问题 multimodal ✅
17 SmartSalesBench: A Resistance-Aware Multi-Turn Benchmark for Evaluating LLM Sales Agents 提出SmartSalesBench以评估抵抗性买家的人格销售代理 large language model
18 Collaborative Theory of Mind: Learning to Self-Revise by Anticipating User Responses 提出协作心智理论以解决用户交互中的自我修正问题 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (15 篇)

#题目一句话要点标签🔗⭐
19 AdaC-GRPO: Towards Mastering Long-Horizon Visual Sudoku Puzzles via Adaptive Curriculum Learning 提出AdaC-GRPO以解决长远视觉数独难题 reinforcement learning curriculum learning large language model
20 RoleWorld: Knowledge Transfer in World Modeling with Functional Role Abstractions 提出RoleWorld以解决世界建模中的知识转移问题 world model world models large language model ✅
21 JANUS: Enhancing Long-Horizon Autonomous Research with Co-Evolving World Models 提出JANUS框架以解决长时间自主研究中的反馈延迟问题 world model world models
22 Mind the Spatial Gap: A Benchmark for Geospatial Reasoning and Constrained Spatial Editing 提出GeoSpatialBench以解决空间推理与编辑能力不足的问题 world model world models large language model
23 Rollout-Time Intervention: Unlocking Agentic Reinforcement Learning Beyond Policy Limits 提出Rollout-Time Intervention以解决强化学习中的策略限制问题 reinforcement learning large language model ✅
24 BoundaryEvo: Data-Model Co-Evolution at the Capability Boundary for Long-Chain Reasoning Distillation 提出BoundaryEvo以解决长链推理蒸馏中的数据选择问题 distillation IMoS
25 A Privileged Teacher That Only Scores Cannot Search: Critique-Conditioned On-Policy Self-Distillation for Reasoning 提出基于批评条件的自蒸馏方法以改善推理能力 distillation privileged information
26 Branching On-Policy Distillation 提出BranchOPD以解决现有OPD方法的监督覆盖不足问题 teacher-student distillation
27 Hindsight Structured Advice Distillation: Steering Executor Agents with Advisors Trained on Their Failures and Task Characteristics 提出后视结构化建议蒸馏以提升执行代理的性能 distillation privileged information
28 Collaborative Learning and Reasoning over Model Slices 提出协作学习以解决单模型推理盲点问题 reinforcement learning distillation large language model
29 Defence against the dark arts of distillation: Unsupervised discovery of subliminal learning 提出无监督方法以发现蒸馏学习中的潜在特征 distillation
30 Beyond Domain-Routed Supervision: Cross-Teacher Guarding for Multi-Teacher On-Policy Distillation 提出跨教师守护机制以解决多教师在线蒸馏中的梯度冲突问题 distillation
31 DP-MIND: Differentially private memory-guided Long context Inference 提出DP-MIND以解决长文本差分隐私推理问题 reinforcement learning large language model
32 Mitigating Capability Degradation in LLM Self-Improvement via Representation Geometry Regularization 提出表示几何正则化以解决LLM自我改进中的能力退化问题 distillation large language model ✅
33 Collective Autonomous Vehicle Routing Needs Social Awareness for Coexistence with Human Drivers 提出社会意识的集体自主车辆路径选择以解决人车共存问题 reinforcement learning penetration

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
34 OGPO: Offline Goal-conditioned Policy Optimization for Recoverable Vision-Language-Action Models 提出OGPO框架以解决VLA模型恢复能力不足问题 manipulation vision-language-action VLA
35 AlphaNIST: Emergent Assembly Skills from Binary Task Rewards 提出AlphaNIST以解决机器人组装任务中的稀疏奖励问题 manipulation

🔬 支柱五：交互与反应 (Interaction & Reaction) (1 篇)

#题目一句话要点标签🔗⭐
36 Humanize Olympiad Agents: A Multi-Session Workflow for Olympiad Problem Solving with Solve, Review, Wiki, Verify 提出HOA框架以提升奥林匹克问题求解能力 IMoS large language model

⬅️ 返回 cs.AI 首页 · 🏠 返回主页