cs.AI（2026-10-08）

📊 共 63 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) (36) 支柱二：RL算法与架构 (RL & Architecture) (22 🔗3) 支柱一：机器人控制 (Robot Control) (4) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (36 篇)

#题目一句话要点标签🔗⭐
1 Recompose and Refine Latent Reasoning Flows for Vision-Language-Action Models 提出FLOWMEM以解决VLA模型中的推理重用问题 vision-language-action VLA multimodal
2 GeoReform: Reflective Formalization Evolution for Multimodal Geometry Problem Solving 提出GeoReform框架以解决多模态几何推理问题 large language model multimodal
3 From Surface to Depth: Towards Cognitive Appraisal Reasoning in Multimodal Emotion Understanding 提出CogEmo框架以解决多模态情感理解中的表层推理问题 large language model multimodal
4 UniData: Universal Multimodal Instruction Generation Pipeline 提出UniData以解决多模态指令生成的挑战 large language model multimodal
5 Internalizer: Portable Context-to-Parameter Mapping for Very Large Language Models 提出Internalizer以实现超大语言模型的上下文到参数映射 large language model
6 Instruction-Conditioned Electromagnetic Spectrum Understanding via Budget-Adaptive Signal Tokenization 提出BATok以解决电磁信号分析中的预算适应性问题 large language model multimodal language conditioned
7 OA-MAP: Evidence-Grounded Multi-Agent Multimodal Framework for Interpretable Knee Osteoarthritis Progression 提出OA-MAP框架以解决膝关节骨关节炎进展预测问题 multimodal
8 From Chain-of-Thought to Loops: Non-Autoregressive Latent Reasoning via Looped Transformers 提出LLoCoT框架以提升语言模型推理效率 chain-of-thought
9 RISR: Residual-Informed Scientific Equation Discovery with Large Language Models 提出残差信息驱动的科学方程发现方法 RISR large language model
10 Can LLMs Fix It Without Code? Toward Automated Verification of No-Code Bug Fixes 提出自动化验证方法以评估无代码修复的有效性 large language model multimodal
11 Overcoming Prior Barriers: Supervised Fine-Tuning under Long-Tail Distribution 提出PASS方法以解决长尾分布下的监督微调问题 large language model
12 Prior or Feedback? What an LLM Uses When Adapting Neural Operators 提出LLM反馈机制以优化神经算子适应性 large language model
13 RouterInterp: Understanding Superposed Specialisation in Mixture of Experts Routing 提出RouterInterp以解决Mixture of Experts模型的可解释性问题 foundation model
14 Evaluating Local Language Model Agents for Reproducible Data Engineering: An Empirical Software Engineering Study of Mobility Workflows 评估本地语言模型代理以支持可重复的数据工程工作 large language model
15 Closed-loop evaluation of LLM agents for embedded software development 提出闭环评估基准以提升嵌入式软件开发中的LLM代理性能 large language model
16 When Lower Reconstruction Loss Hurts: Distributionally Robust Refinement for Low-Bit LLM Quantization 提出分布鲁棒量化以解决低比特LLM量化性能下降问题 large language model
17 Looking Inside LLMs: Small-World Connectivity as a Signature of Reasoning Performance 通过小世界连接性揭示LLM推理性能的结构特征 large language model
18 Use and Disuse: Intent-Structured Experience Consolidation for Memory and Learning in LLM Agents 提出Hippocam以解决大语言模型代理的记忆与学习问题 large language model
19 Is Memorization Context-Sensitive? Prefix-Based Extraction Beyond Isolated Prefixes 研究上下文对语言模型记忆提取的影响 large language model
20 Structure Tax: How Structured Output affects LLMs Performance 提出结构化输出设计以优化大型语言模型性能 large language model
21 MAP4CS: A Multi-dimensional Data Pruning Framework for Efficient Code Retriever Fine-tuning 提出MAP4CS以解决代码检索器微调中的数据冗余问题 large language model
22 Evidence-Traceable Dynamic Interviewer Architecture for Expertise-Adaptive Qualitative Interviews Using Local LLMs 提出证据可追溯的动态访谈架构以解决适应性访谈问题 large language model
23 Where to Adapt Matters: Layer-Selective Fine-Tuning for Capability Retention 提出层选择性微调以解决能力保留问题 large language model
24 AgentEvolver: System-Wide Self-Evolution Through Task Execution 提出AgentEvolver以实现任务执行中的系统自我演化 foundation model
25 Memory Type Varies: Empowering LLM Agents for Long-Term Memory with Diverse Strategies 提出MemoType以解决长时记忆检索策略多样性问题 large language model
26 Tracing the Thoughts of a Coding Agent Playing ARC-AGI-3: Lessons for Continual Learning 提出一种测量协议以追踪编码代理在ARC-AGI-3中的学习过程 foundation model
27 Mine Odyssey: Benchmarking Spatial Agentic Intelligence in the Wild 提出Mine Odyssey基准以评估空间智能代理在真实环境中的表现 foundation model
28 GameCommBench: A Unified Benchmark and Type-Aware Evaluation for AI-Generated Game Commentary 提出GameCommBench以解决AI生成游戏评论的评估问题 multimodal
29 SpireBench: Evaluating Long-Horizon Strategic Play of Agentic Models with Slay the Spire 提出SpireBench以评估智能模型的长期战略游戏能力 large language model
30 Semantic Views: Certified Reuse of LLM Labels Across Semantic Queries 提出SemViews以解决语义查询中的标签重用问题 large language model
31 Faithful and Complete LLM-Assisted Automated Construction of Hierarchical Hardware Knowledge Graphs 提出基于LLM的自动化方法以构建层次化硬件知识图谱 large language model
32 Know What You Don't Know: Agentic Data Synthesis by Evolving Generation Space for Closed-Loop Learning 提出EvoSyn以解决交互代理数据合成中的反馈决策差距问题 large language model
33 Characterizing Overconfident Failure in LLM-Based Code Generation 研究LLM代码生成中的过度自信失败问题 large language model
34 AliO: Output Alignment Matters in Long-Term Time Series Forecasing 提出AliO以解决长时间序列预测中的输出对齐问题 TAMP
35 Language Modeling is Monotone Compression 提出语言建模作为单调压缩以探讨智能与压缩的关系 large language model
36 Intent Graph: Navigating the Analytical Reasoning Space for Exploratory Data Analysis 提出DAG-EDA以解决探索性数据分析中的导航问题 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (22 篇)

#题目一句话要点标签🔗⭐
37 HRIL: Learning Multimodal Synergy via Higher-Order Tensor Modeling 提出HRIL以解决多模态协同信息捕捉问题 representation learning multimodal ✅
38 Learning to Orchestrate Evolutionary Search: Progression-Aware Deep Reinforcement Learning for Dynamic DE-CMA-ES Coordination in Optimization and Structural Model Updating 提出DRL-DCO算法以解决高维结构模型更新问题 reinforcement learning deep reinforcement learning DRL
39 PulseBound: Future-Beat State Forecasting Under an Explicit Information Boundary 提出PulseBound以解决PPG预测中的因果信息访问问题 representation learning MAE PULSE
40 AtomWorld-Mirror: Macro-Step World Modeling of Critical Evolution Backbones for Materials Dynamics 提出AtomWorld-Mirror以解决材料动态模拟中的演化瓶颈问题 world model world models
41 Universal Textual Teaching for LLMs 提出通用文本教学框架以解决知识蒸馏局限性问题 teacher-student distillation implicit representation
42 Constrained Command-Conditioned Reinforcement Learning with Bandit Strategy Selection in Real-Time Strategy Games 提出约束命令条件强化学习以解决实时战略游戏中的策略选择问题 reinforcement learning deep reinforcement learning PPO
43 Environmental Feedback Modeling Matters: Rethinking Feedback Treatment in Agentic Hindsight Self-Distillation 提出SELF框架以优化环境反馈建模在自蒸馏中的应用 reinforcement learning policy learning distillation
44 Safe, Persistent, and Evolving Agent Harness for Understanding Partially Observable Worlds 提出E-Ledger与WorldAbduct以解决部分可观测世界中的安全执行问题 world model world models large language model ✅
45 La-Ribo: RNA Co-Design via Geometry-Latent Flow Matching 提出La-Ribo以解决RNA序列与结构联合设计问题 flow matching
46 When Should Agents Think? Adaptive Reasoning via Cross-Turn Estimation 提出RACE方法以解决智能体推理效率问题 reinforcement learning large language model
47 An Interpretable Approach to PDE Solution Discovery via Structural Experience Distillation 提出SED-MCTS以解决PDE解发现中的可解释性问题 distillation
48 MetaOPD: Meta-Learned Token Weighting for On-Policy Distillation 提出MetaOPD以解决现有蒸馏方法的权重分配问题 distillation
49 A 3D Characterization Framework for Intelligent Sequential Decision Making 提出三维特征框架以优化智能序列决策 reinforcement learning large language model
50 Fed-GRPO: Reward-Signal-Driven Federated Group Relative Policy Optimization 提出Fed-GRPO以解决联邦学习中的隐私问题 reinforcement learning large language model ✅
51 SynCo: Data Synthesis Co-Training for Self-Evolving LLMs via Multi-Agent Reinforcement Learning 提出SynCo框架以解决自我进化LLM的训练数据不匹配问题 reinforcement learning
52 ReCast: Attribution-Oriented Step Representation Learning for LLM-Based Agent Systems 提出ReCast以解决LLM代理系统中的失败归因问题 representation learning
53 MultiWorldBench: Do Independently Controlled Views Describe One Shared World? 提出MultiWorldBench以评估多玩家世界模型的一致性问题 world model world models
54 AuraLuxMuse: Adaptive Fusion Modeling for Aesthetic Stage Lighting Design with Music and Expert Guidance 提出AuraLuxMuse以解决舞台灯光设计的自动化问题 representation learning contrastive learning
55 LLM-IDEA: Identifiability-Driven Experimental Agent for Autonomous Discovery of Mechanistic World Models 提出LLM-IDEA以解决实验可识别性问题 world model world models large language model
56 EvoBelief: Falsifiable World Models for Self-Improving LLM Agents 提出EvoBelief以解决自我演化LLM代理的信念更新问题 world model world models
57 Efficient Audio-Visual Speech Enhancement via Reinforcement Learning-Based Neural Architecture Search 提出基于强化学习的神经架构搜索以实现高效音视频语音增强 reinforcement learning multimodal
58 MultiWorldBench: Do Independently Controlled Views Describe One Shared World? 提出MultiWorldBench以评估多视角一致性问题 world model world models

🔬 支柱一：机器人控制 (Robot Control) (4 篇)

#题目一句话要点标签🔗⭐
59 Cited but Not Consulted: A Counterfactual Audit of Legal Chain-of-Thought Faithfulness 提出法律链条思维的反事实审计方法以提高模型可信度 manipulation large language model chain-of-thought
60 Q-Shaped Options for Hierarchical Reinforcement Learning 提出Q-Shaped Options以解决层次强化学习中的行动抽象问题 locomotion manipulation reinforcement learning
61 EliteTable: Self-Improving Program Search for Executable Control Policies 提出EliteTable以解决可执行控制策略的自我改进问题 bipedal biped large language model
62 Beyond Imitation: A Framework and Benchmark for LLM-Assisted Peer Review 提出基于验证的框架与基准以改善LLM辅助的同行评审 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🔗⭐
63 Mesh-Native Tokenisation: Retrofitting Foundation Models to Unstructured Data 提出Mesh-Native卷积以解决非结构化数据的建模问题 spatiotemporal foundation model

⬅️ 返回 cs.AI 首页 · 🏠 返回主页