cs.AI（2026-05-01）

📊 共 18 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) (9 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (7 🔗2) 支柱一：机器人控制 (Robot Control) (2)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (9 篇)

#题目一句话要点标签🔗⭐
1 EASE: Federated Multimodal Unlearning via Entanglement-Aware Anchor Closure EASE：通过解耦感知锚点闭合实现联邦多模态可遗忘学习 multimodal
2 Empowering Heterogeneous Graph Foundation Models via Decoupled Relation Alignment 提出解耦关系子空间对齐(DRSA)框架，提升异构图基础模型跨域知识迁移能力。 foundation model ✅
3 Can Coding Agents Reproduce Findings in Computational Materials Science? AutoMat：评估LLM智能体在计算材料科学中重现科研结果能力的基准 large language model foundation model
4 LLM-Oriented Information Retrieval: A Denoising-First Perspective 提出面向LLM的信息检索框架，强调去噪以提升检索增强生成质量。 large language model multimodal
5 Social Bias in LLM-Generated Code: Benchmark and Mitigation 提出 Fairness Monitor Agent (FMA) 以缓解 LLM 生成代码中的社会偏见，并提升代码正确性。 large language model chain-of-thought
6 Silicon Showdown: Performance, Efficiency, and Ecosystem Barriers in Consumer-Grade LLM Inference 分析消费级硬件上LLM推理的性能、效率和生态壁垒，揭示Nvidia和Apple Silicon的权衡。 large language model
7 Space Network of Experts: Architecture and Expert Placement 提出Space-XNet框架，解决星载网络中MoE模型的高效分布式部署问题 large language model
8 Skills as Verifiable Artifacts: A Trust Schema and a Biconditional Correctness Criterion for Human-in-the-Loop Agent Runtimes 提出一种基于可验证工件的技能信任模式，用于人机协作Agent运行时环境。 large language model
9 AgentFloor: How Far Up the tool use Ladder Can Small Open-Weight Models Go? AgentFloor：评估小型开源模型在工具使用Agent中能力的阶梯式基准 instruction following

🔬 支柱二：RL算法与架构 (RL & Architecture) (7 篇)

#题目一句话要点标签🔗⭐
10 Physically Native World Models: A Hamiltonian Perspective on Generative World Modeling 提出 Hamiltonian World Models，提升具身智能体物理可靠性和长期预测稳定性。 reinforcement learning world model world models
11 GaMMA: Towards Joint Global-Temporal Music Understanding in Large Multimodal Models GaMMA：面向联合全局-时序音乐理解的大型多模态模型 reinforcement learning multimodal
12 Towards Improving Speaker Distance Estimation through Generative Impulse Response Augmentation 利用生成式脉冲响应增强提升说话人距离估计精度 MAE PULSE
13 Improving LLM Code Generation via Requirement-Aware Curriculum Reinforcement Learning 提出RECRL框架，通过需求感知的课程强化学习提升LLM代码生成能力。 reinforcement learning large language model
14 AEM: Adaptive Entropy Modulation for Multi-Turn Agentic Reinforcement Learning 提出AEM自适应熵调制方法，解决多轮Agent强化学习中的信用分配难题。 reinforcement learning large language model
15 Learn where to Click from Yourself: On-Policy Self-Distillation for GUI Grounding 提出GUI-SD，用于GUI元素定位的On-Policy自蒸馏框架 reinforcement learning distillation ✅
16 DynamicPO: Dynamic Preference Optimization for Recommendation DynamicPO：动态偏好优化，解决LLM推荐系统中负样本过多导致的性能退化问题 DPO direct preference optimization large language model ✅

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
17 Thinking in Text and Images: Interleaved Vision--Language Reasoning Traces for Long-Horizon Robot Manipulation 提出交错视觉-语言推理（IVLR）框架，用于长时程机器人操作任务。 manipulation vision-language-action multimodal
18 Linking Behaviour and Perception to Evaluate Meaningful Human Control over Partially Automated Driving 提出评估部分自动驾驶中人类控制的框架以解决责任与控制的矛盾 shared control

⬅️ 返回 cs.AI 首页 · 🏠 返回主页