cs.AI（2026-11-02）

📊 共 45 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (26 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (16 🔗1) 支柱三：空间感知与语义 (Perception & Semantics) (1) 支柱一：机器人控制 (Robot Control) (1) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (26 篇)

#题目一句话要点标签🔗⭐
1 Same Outputs, Different Scores: Auditing the Unchecked Graders of Multimodal LLM Benchmarks 提出审计多模态LLM基准评分规则以解决评分不一致问题 multimodal
2 Why Tabular Foundation Models Excel on Relational Data They Have Never Seen 提出冻结基础模型以解决关系数据库预测问题 foundation model
3 Variable names provide priors: semantic grounding enables data-efficient numerical prediction by large language models 提出语义基础以提高大语言模型在数值预测中的数据效率 large language model
4 RxnMiner: A Multimodal Agent for Structured Reaction Data Extraction from Chemistry Figures 提出RxnMiner以解决化学文献中反应数据提取问题 multimodal
5 MultiMental-IAS: A Multimodal Driver Mental State Analysis Dataset for Context-Aware Non-IID Evaluation 提出MultiMental-IAS以解决驾驶员心理状态分析问题 multimodal
6 Is your LLM an Agressive Narcissist with Depression? Analyzing Psychological Portraits of Large Language Models 分析大型语言模型的心理特征以提升用户互动体验 large language model
7 Human-Supervised Artificial Intelligence Researchers in Electric Power Systems: Knowledge, Physics, Simulation, and Governance for Large Language Models 提出人类监督的AI研究框架以提升电力系统研究效率 large language model
8 Parallelism-Aligned Pruning and Round-Aware Context Reuse for On-Device Multimodal Agents 提出PARC以解决边缘设备多模态代理的部署不匹配问题 multimodal
9 MAGAR: Multimodal Agents for GUI-Based 3D Mechanical Assembly and Minimal Rework 提出MAGAR框架以解决GUI命令导致的3D机械装配错误问题 multimodal
10 From Natural Language to Exact Cover: A Neuro-Symbolic Approach to Zebra Puzzles 提出神经符号框架以解决自然语言中的斑马逻辑谜题 large language model chain-of-thought
11 Toward Knowledge-Centric AI: Discovering, Internalizing, and Using Explicit Knowledge Across Agents and Generations 提出知识中心AI框架以解决知识传递与内部化问题 large language model foundation model
12 Specification-Aware Predictive Uncertainty Decomposition 提出基于规范的预测不确定性分解以解决信息不足问题 large language model
13 Reimagining computational quantum research with AI scientists 提出OmniQ与RA以解决量子计算研究中的自动化挑战 large language model
14 Chain of Faith: Policy-Internal Rewards for Reasoning Reliance 提出四种奖励机制以增强推理依赖性 chain-of-thought
15 SearchInject: Retrieval Exposure Control for Indirect Prompt Injection in Language-Model Search 提出SearchInject以解决语言模型搜索中的间接提示注入问题 large language model
16 Can Coding Agents Sustain Continual Evolution without Forgetting? 提出持续自我进化的方法以解决编码代理遗忘问题 foundation model ✅
17 Towards a Trustworthy and Governable Internet of Agentics: An Architectural Perspective on Protocols, Systems, and Services 提出互联网代理系统架构以解决可互操作性和治理问题 large language model
18 Math Foundation Formalization Agent: A General-Purpose Approach to Formalizing Mathematical Foundations 提出数学基础形式化代理以解决数学理论自动化问题 large language model
19 Discovering hierarchical governing equations for community-structured complex network dynamics with LLM agents 提出SAGE框架以解决复杂网络动态的方程发现问题 large language model
20 Rethinking Neural Decompilation: Candidate Selection as the Hidden Bottleneck 提出候选选择机制以解决神经反编译中的瓶颈问题 large language model
21 UPSHIFT: User Preference Steering from Human-AI Interaction Feedback Traces 提出UPSHIFT以解决个性化语言模型助手的用户偏好问题 large language model
22 StateFence: Task-State Protection Against Evidence-Chain Contamination in Blue-Team LLM Agents 提出StateFence以解决蓝队LLM代理的证据链污染问题 large language model
23 BasisKV: Value Latents for Communication and Key Routing 提出BasisKV以解决长上下文LLM推理中的KV缓存问题 large language model ✅
24 FORGE: Fault-Oriented Rework Code Generation for Executable Interlocking Assembly 提出FORGE以解决可执行交错装配的故障导向重工代码生成问题 large language model
25 Cumulative and Audited Guardrail Tool Evolution for LLM Agent Safety 提出EVOLVE框架以解决LLM代理安全性问题 large language model
26 Towards Multi-Value Alignment via Value Localization in LLMs 通过价值定位提出多值对齐方法以解决LLMs的价值冲突问题 large language model ✅

🔬 支柱二：RL算法与架构 (RL & Architecture) (16 篇)

#题目一句话要点标签🔗⭐
27 Axion 1.0: Open Agent Foundation Models, Simplified 提出Axion 1.0以简化开放智能体基础模型的训练过程 reinforcement learning DPO direct preference optimization
28 Stage-aware Multimodal Brain Graph Contrastive Learning for Neurodegenerative Disorder Diagnosis 提出阶段感知多模态脑图对比学习以解决神经退行性疾病诊断问题 contrastive learning multimodal
29 Human-Oriented World Model 提出人本世界模型以支持个性化动态预测与状态估计 world model world models
30 SQLoom: Set-Contrastive Learning of Syntactic, Semantic, and Logical Dependencies for Text-to-SQL 提出SQLoom以解决文本到SQL转换中的依赖建模问题 reinforcement learning contrastive learning large language model
31 SpecOPD: Branch-Parallel On-Policy Distillation for Speculative Decoding 提出SpecOPD以解决大语言模型推理速度瓶颈问题 linear attention distillation large language model
32 Multi-Teacher Reasoning Distillation for LLM Recommendation with Self-Exploration 提出MUSE框架以解决推荐系统中推理能力转移问题 contrastive learning distillation large language model
33 RLiD: Reinforcement Learning in the Decision Plane for Language Agents and LLMs 提出RLiD框架以优化语言模型的决策过程 reinforcement learning distillation large language model
34 SGR-OPSD: Learning from Subgoal Plans with Graph-Regularized Credit Assignment for Agentic Reinforcement Learning 提出SGR-OPSD以解决现有OPSD方法的任务进展规划不足问题 reinforcement learning teacher-student distillation
35 Predictive Safety Guardrails Do Not Always Make LLM Agents Safer 揭示预测安全护栏在LLM代理中存在的安全隐患 world model world models large language model
36 Creating Hybrid Interactive Narrative Systems with LLM: From Formal Description to Evaluation 提出基于LLM的混合互动叙事系统以解决叙事管理难题 world model world models large language model
37 From Local Skills to Long-Horizon Web Tasks: Progressive Skill Exploration for LLM Web Agents 提出SCOPE框架以解决长时间Web任务的探索问题 distillation large language model
38 PACERec: Progressive Adaptation of LLM Item Embeddings with Collaborative Evidence for Recommendation 提出PACERec以解决推荐系统中的稀疏交互问题 flow matching large language model
39 BLPO: Beyond-Local Policy Optimization for Long-Horizon LLM Agents 提出BLPO以解决长时间任务中的策略优化问题 reinforcement learning large language model
40 Offline Approximate Value Iteration: A Reinforcement Learning Approach for Satellite Constellation Deployment 提出基于强化学习的近似动态规划方法优化卫星星座部署 reinforcement learning
41 AffToM: Benchmarking Recursive Affective Theory of Mind in Dynamic Multi-Party Interactions 提出AffToM基准以解决动态多方互动中的情感推理问题 reinforcement learning multimodal ✅
42 Spacetime Emergence in the Memory of a Learning Agent 提出在学习智能体记忆中涌现时空结构的方法 reinforcement learning predictive model

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

#题目一句话要点标签🔗⭐
43 Bridging Low-Level Image Fusion and High-Level Semantic Understanding 提出LANCE以解决遥感图像融合与语义理解的脱节问题 open-vocabulary open vocabulary multimodal

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
44 The Customer Never Sees the Goal: Measuring and Detecting Vulnerability-Targeted Manipulation in LLM Assistants 提出GAP方法以检测大型语言模型助手中的操作者诱导操控 manipulation large language model

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
45 Networked Synthetic Social Engineering on the Web 提出网络化合成社会工程框架以应对信息误导问题 ASE

⬅️ 返回 cs.AI 首页 · 🏠 返回主页