cs.AI（2026-02-05）

📊 共 35 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (22 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (10 🔗1) 支柱五：交互与反应 (Interaction & Reaction) (1) 支柱一：机器人控制 (Robot Control) (1) 支柱七：动作重定向 (Motion Retargeting) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (22 篇)

#题目一句话要点标签🔗⭐
1 NEX: Neuron Explore-Exploit Scoring for Label-Free Chain-of-Thought Selection and Model Ranking NEX：基于神经元探索-利用评分的无标签CoT选择与模型排序 large language model chain-of-thought
2 XEmoGPT: An Explainable Multimodal Emotion Recognition Framework with Cue-Level Perception and Reasoning XEmoGPT：提出一种可解释的多模态情感识别框架，关注线索级感知与推理。 multimodal
3 Day-Ahead Electricity Price Forecasting for Volatile Markets Using Foundation Models with Regularization Strategy 提出带正则化策略的基础模型，用于波动市场中的日前电力价格预测。 foundation model
4 A Guide to Large Language Models in Modeling and Simulation: From Core Techniques to Critical Challenges 为建模与仿真应用提供LLM使用指南，强调设计原则、诊断策略和评估方法 large language model
5 A Unified Multimodal Framework for Dataset Construction and Model-Based Diagnosis of Ameloblastoma 提出统一多模态框架，用于成釉细胞瘤数据集构建与模型诊断。 multimodal
6 Clinical Validation of Medical-based Large Language Model Chatbots on Ophthalmic Patient Queries with LLM-based Evaluation 评估医学大语言模型在眼科患者咨询中的表现，并验证基于LLM的评估方法 large language model
7 Position: Universal Time Series Foundation Models Rest on a Category Error 质疑时间序列通用基础模型，提出因果控制代理范式以提升泛化能力。 foundation model
8 Hallucination-Resistant Security Planning with a Large Language Model 提出一种抗幻觉的安全规划框架，利用大语言模型提升事件响应效率。 large language model
9 Surgery: Mitigating Harmful Fine-Tuning for Large Language Models via Attention Sink 提出Surgery，通过注意力Sink机制缓解大语言模型有害微调带来的安全风险。 large language model ✅
10 Exploring AI-Augmented Sensemaking of Patient-Generated Health Data: A Mixed-Method Study with Healthcare Professionals in Cardiac Risk Reduction 利用AI增强患者生成健康数据的理解：心脏风险降低中医疗专业人员的混合方法研究 large language model multimodal
11 AgenticPay: A Multi-Agent LLM Negotiation System for Buyer-Seller Transactions AgenticPay：一个用于买卖交易的多智能体LLM协商系统 large language model ✅
12 Split Personality Training: Revealing Latent Knowledge Through Alternate Personalities 提出分裂人格训练SPT，通过交替人格揭示大语言模型中的潜在知识 large language model
13 DyTopo: Dynamic Topology Routing for Multi-Agent Reasoning via Semantic Matching DyTopo：基于语义匹配的动态拓扑多智能体推理框架 large language model
14 Compound Deception in Elite Peer Review: A Failure Mode Taxonomy of 100 Fabricated Citations at NeurIPS 2025 揭示AI生成文献引用欺骗：NeurIPS 2025中100个伪造引用的失效模式分析 large language model
15 Towards Green AI: Decoding the Energy of LLM Inference in Software Development 分析LLM推理能耗，提出抑制“胡言乱语”行为以降低软件开发能耗 large language model
16 Determining Energy Efficiency Sweet Spots in Production LLM Inference 提出Transformer架构能耗分析模型，优化LLM推理能效 large language model
17 Graph-based Agent Memory: Taxonomy, Techniques, and Applications 综述：基于图结构的Agent记忆，实现知识积累、迭代推理和自我进化。 large language model ✅
18 Generative Ontology: When Structured Knowledge Learns to Create 提出生成式本体框架，融合本体知识与大语言模型创造力，实现结构化内容生成。 large language model
19 Capture the Flags: Family-Based Evaluation of Agentic LLMs via Semantics-Preserving Transformations 提出Evolve-CTF，通过语义保持变换评估Agentic LLM在CTF任务中的鲁棒性。 large language model
20 SDFP: Speculative Decoding with FIT-Pruned Models for Training-Free and Plug-and-Play LLM Acceleration SDFP：基于FIT剪枝模型的免训练推测解码，加速LLM推理。 large language model
21 RaBiT: Residual-Aware Binarization Training for Accurate and Efficient LLMs 提出RaBiT以解决大语言模型量化中的特征共适应问题 large language model
22 EGSS: Entropy-guided Stepwise Scaling for Reliable Software Engineering 提出熵引导的逐步扩展（EGSS）框架，提升软件工程任务性能并降低计算开销。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (10 篇)

#题目一句话要点标签🔗⭐
23 RL-VLA$^3$: Reinforcement Learning VLA Accelerating via Full Asynchronism 提出RL-VLA$^3$，通过全异步加速VLA模型的强化学习训练。 reinforcement learning vision-language-action VLA
24 LMMRec: LLM-driven Motivation-aware Multimodal Recommendation LMMRec：提出基于LLM的动机感知多模态推荐框架，提升推荐性能。 contrastive learning large language model multimodal
25 TKG-Thinker: Towards Dynamic Reasoning over Temporal Knowledge Graphs via Agentic Reinforcement Learning 提出TKG-Thinker，通过Agent强化学习实现时序知识图谱的动态推理 reinforcement learning large language model
26 ProAct: Agentic Lookahead in Interactive Environments ProAct：通过Agent内部前瞻推理提升交互环境中LLM智能体的规划能力 PPO distillation large language model ✅
27 Refine and Purify: Orthogonal Basis Optimization with Null-Space Denoising for Conditional Representation Learning 提出OD-CRL框架，通过正交基优化与零空间去噪提升条件表示学习性能 representation learning
28 Total Variation Rates for Riemannian Flow Matching 提出非渐近总变差收敛分析以优化流匹配算法 flow matching
29 Quantum Reinforcement Learning with Transformers for the Capacitated Vehicle Routing Problem 提出基于Transformer的量子强化学习方法，解决带容量约束车辆路径问题 reinforcement learning
30 Reasoning-guided Collaborative Filtering with Language Models for Explainable Recommendation RGCF-XRec：融合推理引导的协同过滤与语言模型，实现可解释推荐 representation learning large language model
31 ALIVE: Awakening LLM Reasoning via Adversarial Learning and Instructive Verbal Evaluation 提出ALIVE框架以解决大语言模型推理能力瓶颈问题 reinforcement learning large language model
32 AgentXRay: White-Boxing Agentic Systems via Workflow Reconstruction AgentXRay：通过工作流重构实现Agentic系统的白盒化 distillation large language model

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

#题目一句话要点标签🔗⭐
33 FHAIM: Fully Homomorphic AIM For Private Synthetic Data Generation 提出FHAIM，利用全同态加密实现隐私保护的合成数据生成。 OMOMO

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
34 Agent2Agent Threats in Safety-Critical LLM Assistants: A Human-Centric Taxonomy 提出AgentHeLLM框架，应对LLM智能座舱Agent2Agent通信中的安全威胁 manipulation large language model

🔬 支柱七：动作重定向 (Motion Retargeting) (1 篇)

#题目一句话要点标签🔗⭐
35 TangramSR: Can Vision-Language Models Reason in Continuous Geometric Space? TangramSR：提出基于视觉-语言模型的切磋拼图自精炼框架，提升连续几何空间推理能力 geometric consistency

⬅️ 返回 cs.AI 首页 · 🏠 返回主页