cs.AI（2026-08-03）

📊 共 40 篇论文 | 🔗 2 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (18 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (17 🔗1) 支柱三：空间感知与语义 (Perception & Semantics) (4) 支柱一：机器人控制 (Robot Control) (1)

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

#题目一句话要点标签🔗⭐
1 Exploring and Bridging Knowledge Holes in Unlearned Multimodal Large Language Models 提出选择性保护与锚定正则化以解决多模态大语言模型的知识缺失问题 large language model multimodal
2 Can Foundation Models Hear What Made That Sound? A Tiered Benchmark of Audio-Language Models and Traditional Classifiers for Closed-Set Sound Source Identification 提出音频语言模型的分层基准以解决声音源识别问题 foundation model chain-of-thought
3 MonitrLLM: A Community-Centered Evaluation Infrastructure for Large Language Models 提出MonitrLLM以解决LLM评估中用户反馈缺失的问题 large language model
4 Deferred Exposure of Future Trajectories for Verifiable Reasoning in Autonomous Driving VLMs 提出DEFT-RLVR以解决自主驾驶推理中的轨迹偏差问题 vision-language-action VLA chain-of-thought
5 Emergence Invariance: From Symbolized Thought to Interface Refinement 提出符号化思维与界面优化的紧密联系以解决认知缺陷问题 large language model chain-of-thought
6 Right Answer, Wrong Method: Shortcut Hacking Misleads the Evaluation of LLM Reasoning on Frontier Science Benchmarks 提出反作弊策略以解决大型语言模型推理评估问题 large language model
7 SkillTrace: Traversing a Query-Skill Graph for Composable LLM Agents 提出SkillTrace以解决可组合LLM代理的技能检索问题 large language model
8 MemArbiter: Decision-Time Memory Arbitration for Long-Horizon LLM Agents 提出MemArbiter以解决长时间任务中的记忆管理问题 large language model
9 EduZone: A Framework for Evaluating LLM Safety for K-12 Students and Teachers 提出EduZone框架以评估K-12教育中LLM的安全性问题 large language model
10 Energy-Efficient LLM Serving via Disaggregated Attention--FFN and Flexible Frequency Scaling 提出AFlex框架以解决大语言模型服务中的能效问题 large language model
11 FOCUS: FP4 Optimization via Coupled-Relaxation and Dual-Granularity Scaling 提出FOCUS框架以优化FP4量化精度问题 large language model ✅
12 PICopilot: An LLM-based Agentic Framework for Assisting Photonic Integrated Circuit Design via Script Generation 提出PICopilot以解决光子集成电路设计中的脚本生成问题 large language model
13 MNC: Scope-Bound Semantic Declassification for Private LLM-Agent Communication 提出MNC以解决多代理大语言模型的隐私泄露问题 large language model
14 Coding Agents as Test-Suite Auditors: Finding What Official Suites Miss While Approaching What They Catch 提出编码代理作为测试套件审计工具以解决官方套件遗漏问题 large language model
15 Constructing Executable Analytical Knowledge Representations for Meta-Analysis Synthesis Using an Agentic Harness 提出可执行分析知识表示以解决元分析合成问题 large language model
16 Beyond Single-Use Tokens: Durable Authorization State for Replay-Resistant LLM Agent Actions 提出CapLease以解决LLM代理的重放抵抗授权问题 large language model
17 Allocation Before Ranking: Decoupled Token Compression for OmniLLMs 提出Macer以解决OmniLLMs中的令牌压缩问题 multimodal
18 GISAgentBench: A Practitioner-Sourced Benchmark for Evaluating LLM Agents on GIS Tasks 提出GISAgentBench以解决GIS任务评估的不足问题 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (17 篇)

#题目一句话要点标签🔗⭐
19 Faster-WAM: Do World Action Models Need Deep Action Modules? 提出Faster-WAM以解决现有世界动作模型的计算开销问题 world model world models world action model
20 Instruction-Conditioned Exploration with Asymmetric Reinforcement Learning and Self-Distillation 提出指令条件探索以解决大语言模型的探索问题 reinforcement learning distillation large language model
21 PCSD: Persistent Consistency for Self-Distillation in Agentic Reinforcement Learning 提出PCSD以解决强化学习中的稀疏奖励问题 reinforcement learning distillation large language model
22 Antares: Foundation Models for Agentic Vulnerability Localization 提出Antares以解决软件安全中的漏洞定位问题 reinforcement learning foundation model
23 ProWorld: Progress-Aware Hyperbolic World Models for Long-Horizon Visual Goal Reaching 提出ProWorld以解决长时间视觉目标规划中的进展问题 world model world models JEPA
24 Hear, Invoke, and Understand: A Skill-Calling Multimodal Agent for Large Audio Language Models 提出SpeechAgent-R以解决复杂音频推理问题 reinforcement learning multimodal
25 Chess on Ice: Curling Tactical Decision-Making via Backward Induction and Deep Reinforcement Learning 提出深度强化学习框架以解决冰壶战术决策问题 reinforcement learning deep reinforcement learning
26 CoNav-UAV: Cooperative Dual-Altitude Aerial Navigation via Stackelberg Learning 提出CoNav-UAV以解决无人机协同导航问题 distillation privileged information VLN
27 Mamba with Hierarchical Memory: Solving Representation Bottleneck in Long Sequence Modeling 提出层次记忆Mamba以解决长序列建模中的表示瓶颈问题 Mamba linear attention
28 DAPD: Dual-Anchored Policy Distillation 提出双锚政策蒸馏以解决信息不对称问题 distillation privileged information
29 Is More Privileged Information Better? From Solution Traces to Problem-Solving Structure in Self-Distilled Reasoning 提出PS-OPSD以提升自蒸馏推理的准确性 distillation privileged information
30 RL-Lock: Reinforcement Learning for Generating Interlocking Assemblies 提出RL-Lock以解决生成互锁装配问题 reinforcement learning
31 Agentic Incident Response through Digital Twin-Enhanced Multiscale Planning 提出基于数字双胞胎的多尺度规划以优化事件响应 reinforcement learning large language model
32 Harness-R1: Learning to Edit Executable Runtime Harnesses from Agent Failure Trajectories 提出Harness-R1以实现可执行运行时的智能编辑 reinforcement learning large language model
33 Beyond the Mean: Multi-Moment Policy Optimization for LLM Reasoning 提出多时刻策略优化方法以提升大语言模型推理能力 reinforcement learning large language model
34 CoEvoKG: Co-Evolving Knowledge Graphs with Self-Evolving Search Agents 提出CoEvoKG框架以解决自我演化搜索代理知识积累不足问题 reinforcement learning large language model ✅
35 Syntax Meets Semantics: Understanding Scientific Formulae 提出跨模态对齐方法以提升科学公式检索性能 representation learning contrastive learning

🔬 支柱三：空间感知与语义 (Perception & Semantics) (4 篇)

#题目一句话要点标签🔗⭐
36 DeGS: A Scalable 3DGS Architecture via Decoupled Workload Parsing and Reorganization 提出DeGS架构以解决3DGS加速器的可扩展性问题 3D gaussian splatting 3DGS gaussian splatting
37 Abduction Without a Body? Representational Grounding and the Abduction Loop for Scientific Hypothesis Generation 提出一种新架构以解决科学假设生成中的身份推理问题 affordance multimodal
38 Long-Horizon Autonomous Architecture Research with a Language-Model Agent: A Behavioural Case Study 利用语言模型代理进行长时间自主架构研究 affordance large language model
39 Rethinking Generative AI Literacy: An Integrative, Developmental, and Dialectical Framework for K-12 Teacher Education 提出RAIL-Ed框架以解决K-12教师生成AI素养不足问题 affordance large language model

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
40 Securing Agentic AI: From Per-Action Checks to Trajectory Assurance 提出行为轨迹保障机制以解决自主智能体安全问题 manipulation large language model

⬅️ 返回 cs.AI 首页 · 🏠 返回主页