cs.AI（2026-05-26）

📊 共 35 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (23) 支柱二：RL算法与架构 (RL & Architecture) (6) 支柱四：生成式动作 (Generative Motion) (2 🔗1) 支柱五：交互与反应 (Interaction & Reaction) (1) 支柱三：空间感知与语义 (Perception & Semantics) (1) 支柱八：物理动画 (Physics-based Animation) (1) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (23 篇)

#题目一句话要点标签🔗⭐
1 Reasoning, Code, or Both? How Large Language Models Handle Variations in Math Questions 对比思维链、单步代码执行与迭代代码执行，评估大语言模型在数学问题变体上的鲁棒性。 large language model chain-of-thought
2 Boosting Knowledge Graph Foundation Models via Enhanced Negative Sampling 提出KMAS自适应负采样方法，提升知识图谱基础模型在零样本知识图谱补全任务上的性能。 foundation model
3 Generating Robust Portfolios of Optimization Models using Large Language Models 利用大语言模型生成优化模型组合，提升决策鲁棒性 large language model
4 What Makes Chain-of-Thought Work at Probe Time? Local Co-occurrence Rather Than Global Derivation 提出局部共现激活模型以解析链式思维的有效性 chain-of-thought
5 LiveK12Bench: Have Large Multimodal Models Truly Conquered High School-level Examinations? 提出LiveK12Bench，评估大型多模态模型在真实高中考试场景下的推理能力 multimodal
6 Beyond a Single Direction: Chain-of-Thought Disrupts Simple Steering of Refusal 思维链干扰拒绝行为的简单引导：揭示大型推理模型的新型攻击面 chain-of-thought
7 MedGuideX: Internalizing Decision Logic from Executable Guidelines into Large Language Models for Clinical Reasoning MedGuideX：将可执行指南的决策逻辑融入大型语言模型，用于临床推理。 large language model
8 Gumbel Machine: Counterfactual Student Writing Generation via Gumbel Noise Steering Gumbel Machine：通过Gumbel噪声引导生成反事实学生写作文本 large language model instruction following
9 MUSE-Autoskill: Self-Evolving Agents via Skill Creation, Memory, Management, and Evaluation MUSE-Autoskill：通过技能生命周期管理实现自进化Agent large language model
10 Cordyceps: Covert Control Attacks on LLMs via Data Poisoning Cordyceps：通过数据投毒对LLM进行隐蔽控制攻击 large language model
11 GENESIS: Harnessing AI Agents for Autonomous 6G RAN Synthesis, Research, and Testing GENESIS：利用AI Agent实现6G RAN的自主合成、研究与测试 large language model
12 Qiskit QuantumKatas: Adapting Microsoft's Quantum Computing exercises for LLM evaluation 构建Qiskit QuantumKatas基准，用于评估LLM在量子计算任务中的能力。 chain-of-thought
13 Learning to Act under Noise: Enhancing Agent Robustness via Noisy Environments NoisyAgent：通过噪声环境训练提升LLM智能体在真实场景下的鲁棒性 large language model
14 VitaBench 2.0: Evaluating Personalized and Proactive Agents in Long-Term User Interactions VitaBench 2.0：评估长期用户交互中个性化和主动型Agent large language model
15 Traceable Knowledge Graph Reasoning Enables LLM-Assisted Decision Support for Industrial VOCs in the Steel Industry Chat-ISV：基于可溯源知识图谱推理的钢铁行业VOCs治理LLM辅助决策系统 large language model
16 ConVer: Using Contracts and Loop Invariant Synthesis for Scalable Formal Software Verification ConVer：利用合约与循环不变式综合实现可扩展的形式化软件验证 large language model
17 ReasonOps: A Unified Operational Paradigm for Trustworthy Verified LLM Reasoning 提出ReasonOps：一种可信、可验证的大语言模型推理统一操作范式 large language model
18 Strategies for Guiding LLMs to Use Software Design Patterns: A Case of Singleton 探索引导LLM应用Singleton设计模式的策略，提升代码质量与一致性 large language model
19 Persistent AI Agents in Academic Research: A Single-Investigator Implementation Case Study 构建持久化AI Agent科研环境，探索其在学术研究中的应用与性能 large language model
20 MatFormBench: A Benchmarking Evaluation Framework for Target-Driven Materials Formulation MatFormBench：针对目标驱动材料配方设计的综合性基准测试框架 large language model
21 Towards Feedback-to-Plan Decisions for Self-Evolving LLM Agents in CUDA Kernel Generation 提出CUDAnalyst，用于分析LLM智能体在CUDA核生成中反馈到规划决策的影响。 large language model
22 L2Rec: Towards Dual-View Understanding of LLMs for Personalized Recommendation L2Rec：通过双视角理解LLM，实现个性化推荐 large language model
23 Plans for Evaluating Structured Generative Search Summaries 提出评估结构化生成式搜索摘要的框架，用于提升网络搜索结果的呈现效果。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (6 篇)

#题目一句话要点标签🔗⭐
24 PolyFusionAgent: A Multimodal Foundation Model and Autonomous AI Assistant for Polymer Property Prediction and Inverse Design PolyFusionAgent：用于聚合物性质预测和逆向设计的交互式多模态基础模型与自主AI助手 representation learning foundation model multimodal
25 Alignment Tampering: How Reinforcement Learning from Human Feedback Is Exploited to Optimize Misaligned Biases 揭示RLHF对齐中存在的篡改漏洞，可被LLM利用以放大偏差。 reinforcement learning RLHF large language model
26 StepOPSD: Step-Aware Online Preference Distillation for Agent Reinforcement Learning 提出StepOPSD：一种步感知的在线偏好蒸馏方法，提升Agent强化学习的局部决策能力 reinforcement learning distillation
27 StreamSplit: Continuous Audio Representation Learning via Uncertainty-Guided Adaptive Splitting StreamSplit：通过不确定性引导的自适应分割实现连续音频表征学习 reinforcement learning representation learning contrastive learning
28 Counteraction-Aware Multi-Teacher On-Policy Distillation for General Capability Recovery with Domain Preservation 提出CaMOPD，通过对抗解耦和差距采样，提升领域模型通用能力恢复效果。 teacher-student distillation
29 UnityMAS-O: A General RL Optimization Framework for LLM-Based Multi-Agent Systems UnityMAS-O：用于LLM多智能体系统的通用强化学习优化框架 reinforcement learning PPO

🔬 支柱四：生成式动作 (Generative Motion) (2 篇)

#题目一句话要点标签🔗⭐
30 PilotTTS: A Disciplined Modular Recipe for Competitive Speech Synthesis PilotTTS：通过精简架构和严格数据工程实现高质量语音合成 motion synthesis ✅
31 Lessons from Penetration Tests on Large-Scale Agent Systems 大规模Agent系统渗透测试揭示的安全漏洞及改进措施 penetration

🔬 支柱五：交互与反应 (Interaction & Reaction) (1 篇)

#题目一句话要点标签🔗⭐
32 Practical Anonymous Two-Party Gradient Boosting Decision Tree 提出匿名两方梯度提升决策树训练方法，解决ID泄露问题，兼顾效率与隐私。 OMOMO

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

#题目一句话要点标签🔗⭐
33 The Sensation Modulating Network:Haltability as the architectural ground for object-directed phenomenology 提出感觉调节网络（SMN），为具身认知架构提供了一种新的解决方案。 affordance

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

#题目一句话要点标签🔗⭐
34 Structure-Adaptive Conformal Inference for Large-Scale Out-of-Distribution Testing 提出结构自适应共形推断方法，用于大规模分布外测试。 spatiotemporal

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
35 From Static Context to Calibrated Interactive RL: Mitigating Distribution Shift in Multi-turn Dialogue with Aligned Simulator 提出校准交互式强化学习，缓解多轮对话中由分布偏移导致的问题。 sim-to-real

⬅️ 返回 cs.AI 首页 · 🏠 返回主页