cs.AI（2025-09-29）

📊 共 28 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) (19 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (8) 支柱三：空间感知与语义 (Perception & Semantics) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (19 篇)

#题目一句话要点标签🔗⭐
1 Building the EHR Foundation Model via Next Event Prediction 提出基于事件预测的EHR基础模型，提升LLM在临床时序推理能力 large language model foundation model
2 Radiology's Last Exam (RadLE): Benchmarking Frontier Multimodal AI Against Human Experts and a Taxonomy of Visual Reasoning Errors in Radiology RadLE：放射学专家级诊断基准，评估多模态AI并分析视觉推理错误 large language model multimodal
3 TimeOmni-1: Incentivizing Complex Reasoning with Time Series in Large Language Models 提出TimeOmni-1，通过时间序列推理套件TSR-Suite，解决大语言模型在复杂时间序列推理任务中的挑战。 large language model multimodal
4 Bridging the behavior-neural gap: A multimodal AI reveals the brain's geometry of emotion more accurately than human self-reports 多模态AI超越人类自报告，更准确揭示大脑情感几何 large language model multimodal ✅
5 Evaluating Foundation Models with Pathological Concept Learning for Kidney Cancer 利用病理概念学习评估肾癌Foundation Model的转化能力 foundation model ✅
6 AdvChain: Adversarial Chain-of-Thought Tuning for Robust Safety Alignment of Large Reasoning Models AdvChain：对抗性思维链微调，提升大型推理模型安全对齐的鲁棒性 chain-of-thought
7 MGM-Omni: Scaling Omni LLMs to Personalized Long-Horizon Speech MGM-Omni：面向个性化长时程语音的通用多模态大语言模型 multimodal
8 Causal Autoencoder-like Generation of Feedback Fuzzy Cognitive Maps with an LLM Agent 提出基于LLM的反馈模糊认知图自编码器，实现可解释的因果关系学习。 large language model
9 ATLAS: Constraints-Aware Multi-Agent Collaboration for Real-World Travel Planning ATLAS：面向真实旅行规划的约束感知多智能体协作框架 large language model
10 A(I)nimism: Re-enchanting the World Through AI-Mediated Object Interaction A(I)nimism：通过AI中介的物体交互，探索万物有灵论的新可能 large language model
11 Toxicity in Online Platforms and AI Systems: A Survey of Needs, Challenges, Mitigations, and Future Directions 提出在线平台和AI系统中内容毒性的综合分类，并探讨检测与缓解策略。 large language model
12 Adaptive Test-Time Reasoning via Reward-Guided Dual-Phase Search 提出基于奖励引导的双阶段搜索，提升LLM在推理任务中的效率与准确性。 large language model
13 AutoCode: LLMs as Problem Setters for Competitive Programming AutoCode：利用大型语言模型自动生成竞赛级编程题目 large language model
14 ReasoningBank: Scaling Agent Self-Evolving with Reasoning Memory 提出 ReasoningBank，通过推理记忆和自进化提升Agent在持续任务中的性能。 large language model
15 Dive into the Agent Matrix: A Realistic Evaluation of Self-Replication Risk in LLM Agents 提出LLM Agent自复制风险评估框架，揭示实际应用中潜在安全隐患 large language model
16 Flash-Searcher: Fast and Effective Web Agents via DAG-Based Parallel Execution Flash-Searcher：基于DAG并行执行的快速高效Web Agent large language model
17 MASLegalBench: Benchmarking Multi-Agent Systems in Deductive Legal Reasoning 提出MASLegalBench：用于评估多智能体系统在演绎法律推理中的性能 large language model
18 Neural network embeddings recover value dimensions from psychometric survey items on par with human data 提出SQuID方法，利用神经网络嵌入从心理测量问卷条目中恢复价值维度，性能与人类数据相当。 large language model
19 Experience-Guided Reflective Co-Evolution of Prompts and Heuristics for Automatic Algorithm Design 提出EvoPH框架，通过经验引导的提示与启发式算法协同进化，实现自动算法设计 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (8 篇)

#题目一句话要点标签🔗⭐
20 Training Agents Inside of Scalable World Models Dreamer 4：通过可扩展世界模型在Minecraft中实现离线钻石获取 reinforcement learning world model dreamer
21 Hybrid Reward Normalization for Process-supervised Non-verifiable Agentic Tasks 提出PPR方法，通过混合奖励归一化提升Agent在非验证任务中的表现 reinforcement learning large language model
22 Modeling Others' Minds as Code ROTE：利用程序合成高效预测人类及AI行为，提升人机协作 behavior cloning large language model
23 DeepSearch: Overcome the Bottleneck of Reinforcement Learning with Verifiable Rewards via Monte Carlo Tree Search DeepSearch：通过蒙特卡洛树搜索和可验证奖励克服强化学习瓶颈 reinforcement learning
24 The Era of Real-World Human Interaction: RL from User Conversations 提出基于用户对话的强化学习(RLHI)，实现个性化对齐和持续模型改进。 reinforcement learning instruction following
25 Pushing LLMs to Their Logical Reasoning Bound: The Role of Data Reasoning Intensity 提出数据推理强度（DRI）指标，优化训练数据以提升LLM逻辑推理能力。 reinforcement learning large language model
26 Towards Safe Reasoning in Large Reasoning Models via Corrective Intervention 提出Intervened Preference Optimization以提升大型推理模型安全性 preference learning chain-of-thought
27 Learning to Interact in World Latent for Team Coordination 提出交互世界隐空间(IWoL)框架，促进多智能体强化学习中的团队协作 reinforcement learning representation learning

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

#题目一句话要点标签🔗⭐
28 Vision-and-Language Navigation with Analogical Textual Descriptions in LLMs 提出基于LLM中类比文本描述的视觉-语言导航方法，提升导航性能。 scene understanding embodied AI VLN

⬅️ 返回 cs.AI 首页 · 🏠 返回主页