cs.AI（2026-09-30）

📊 共 37 篇论文 | 🔗 3 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (21 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (13 🔗2) 支柱一：机器人控制 (Robot Control) (2) 支柱三：空间感知与语义 (Perception & Semantics) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (21 篇)

#题目一句话要点标签🔗⭐
1 WorldAuditBench: Interactive 3D World Auditing with Multimodal Agents 提出WorldAuditBench以解决3D世界审计中的异常检测问题 vision-language-action VLA multimodal
2 CAMOS: Coupled Oscillatory State-Space Model for Multimodal Clinical Time-Series 提出CAMOS以解决多模态临床时间序列数据缺失问题 multimodal zero-shot transfer
3 Grounding Time-Series Foundation Models in Digital Twin Topology for Predictive Maintenance 提出拓扑信息融合方法以提升数字双胞胎中的时间序列预测性能 foundation model
4 CodeMimicry: Exploiting Safety Generalization Lag in Large Language Models via Structured Code Completion 提出CodeMimicry以解决大型语言模型安全对齐滞后问题 large language model
5 Who Owns That? Evaluating Ownership Intuitions in Large Language Models 提出竞争所有权归属任务以评估大型语言模型的所有权直觉 large language model
6 MASCRDM: Multi-Agent System for Compliance Risk Detection and Mitigation in Training Process of Large Language Models 提出MASCRDM以解决大语言模型合规风险检测与缓解问题 large language model
7 Targeted Retrieval, Compact Representations: How CoT Reasoning Improves Long-Context Counting 提出目标检索与紧凑表示以提升长文本计数精度 large language model chain-of-thought
8 BayesNDE: Bayesian Generative Modeling for Neural Density Estimation 提出BayesNDE以解决神经密度估计问题 multimodal ✅
9 Can Computation from Earlier Problems Help LLMs Solve New Ones? 提出STAIR以提升大语言模型在新问题上的解决能力 large language model
10 How Much of a Harness Does a Strong Agent Need for Autonomous ML Engineering? 提出简化编码代理的架构以优化自主机器学习工程 large language model
11 FIGS: Evaluating Multi-Turn Sycophancy Without Penalizing Empathy 提出FIGS框架以解决多轮对话中的谄媚与同理心平衡问题 large language model
12 Self-Spec Verifiable Code Generation 提出VeriCodeBench以解决LLM生成代码的可靠性问题 large language model
13 Speculative Safety Honeypot: Toward Proactive Defense Against Multi-turn Agent Attacks 提出Speculative Safety Honeypot以解决多轮代理攻击问题 large language model
14 Inferring Causal Relations between Two Sequences of Events with Language Models 利用大语言模型推断事件序列间的因果关系 large language model
15 WorkGenesis: Building the Worlds That Teach Agents to Work 提出WorkGenesis以构建可执行的职业工作场景 large language model
16 MiniRep: Robust Reputation-Based Aggregation for Multi-Agent Debate 提出MiniRep以解决多智能体辩论中的信誉评估问题 large language model
17 ANI: Adaptive Numerical Injection for Unifying Semantic and Arithmetic Representations in Numerical Reasoning 提出ANI以解决大语言模型中的数值推理问题 large language model
18 EngramBench: A Capability-Grounded Benchmark for Skill-Evolution Harnesses 提出EngramBench以解决自主体技能演化评估中的识别问题 large language model
19 Faithful Dual-constrained Erasure for Robust LLM Safety Alignment 提出FDCU框架以解决大语言模型的安全对齐问题 large language model
20 Rep2Skill: Representation-Guided Skill Self-Evolution for LLM Agents 提出Rep2Skill以解决LLM代理技能自我演化问题 large language model
21 Whose Voice Survives the Summary? A Voice-Retention Audit of LLM Employee Listening 提出语音保留/表现比率以解决LLM总结中的偏见问题 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (13 篇)

#题目一句话要点标签🔗⭐
22 Reinforcing Multimodal Reasoning via Token-Level Perception-Grounded Advantage Estimation 提出基于令牌级感知的优势估计以提升多模态推理能力 reinforcement learning large language model multimodal ✅
23 UniEvo-VL: An On-policy Self-Distillation Training Recipe for Multimodal Model Self-improvement 提出UniEvo-VL以实现多模态模型的自我改进 distillation privileged information multimodal
24 Beyond Prediction: Steering VLM Agents with Retrospective World Modeling 提出回溯世界建模以提升VLM代理的决策能力 reinforcement learning policy learning world model
25 AVERT-VLN: Abstention-aware Visual Error Recovery and Training for Vision-and-Language Navigation 提出AVERT-VLN以解决视觉语言导航中的错误恢复问题 preference learning VLN
26 Why Do Conventional World Models Fail to Learn Cellular Automata? 提出改进方案以解决传统世界模型无法学习细胞自动机的问题 world model world models
27 SkillFM: Generating Skills for LLM Agents via Latent Flow Matching 提出SkillFM框架以生成LLM代理的文本技能 reinforcement learning flow matching large language model ✅
28 GrammarRL: Effective Grammar-Constrained Decoding via Reinforcement Learning 提出GrammarRL以解决语法约束生成中的语义质量问题 reinforcement learning instruction following
29 Completion-Aware Cross-Fidelity Offline-to-Online Reinforcement Learning for Multi-Line Bus Holding 提出跨信度离线到在线强化学习以解决多线路公交调度问题 reinforcement learning
30 Learning Steganography Is Easy, Learning Steganographic Reasoning Is Hard 研究揭示隐写推理学习的困难性及其与隐写消息的关系 reinforcement learning chain-of-thought
31 Disentangling Self-Distillation: Measuring and Modeling Acquisition and Retention 提出统一框架以优化自蒸馏过程中的获取与保留问题 distillation
32 Multi-LLM Collaborative Alignment via Stackelberg Games 提出Stackelberg Alignment以解决多语言模型协作学习问题 DPO instruction following
33 MeanVoiceFlow2: Joint Optimization of Mean Flow and Content Encoder for Fast One-Step Zero-Shot Voice Conversion 提出MeanVoiceFlow2以解决语音转换中的计算瓶颈问题 flow matching distillation
34 OverForge: Reasoning Through Strategies and Tactics Helps Cooperative Lifelong Adaptation 提出OverForge以解决协作学习中的策略与战术适应问题 world model world models

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
35 ChronoGraph: Functional 4D Scene Graphs with Vision-Language Models for Interaction Understanding and Grounded Planning 提出ChronoGraph以解决4D交互理解与规划问题 manipulation mobile manipulation reinforcement learning
36 AIMS: An Agentic AI Framework for Sim-to-Real Multi-Modal ISAC 提出AIMS框架以解决多模态ISAC的仿真到现实问题 sim-to-real scene understanding

🔬 支柱三：空间感知与语义 (Perception & Semantics) (1 篇)

#题目一句话要点标签🔗⭐
37 Belief-Aware Multi-Agent Path Finding under Map Uncertainty 提出MAGIC框架以解决MAPF中的地图不确定性问题 traversability

⬅️ 返回 cs.AI 首页 · 🏠 返回主页