cs.AI（2026-01-22）

📊 共 23 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) (17 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (5) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (17 篇)

#题目一句话要点标签🔗⭐
1 Multimodal Climate Disinformation Detection: Integrating Vision-Language Models with External Knowledge Sources 结合外部知识的视觉-语言模型用于多模态气候虚假信息检测 multimodal
2 Grounding Large Language Models in Reaction Knowledge Graphs for Synthesis Retrieval 提出基于反应知识图谱的LLM合成检索方法，解决化学合成规划中LLM的幻觉问题。 large language model ✅
3 THOR: A Versatile Foundation Model for Earth Observation Climate and Society Applications 提出THOR，一种适用于地球观测气候和社会应用的多功能基础模型 foundation model
4 ErrorMap and ErrorAtlas: Charting the Failure Landscape of Large Language Models 提出ErrorMap和ErrorAtlas，用于剖析大语言模型失败原因并构建错误分类体系。 large language model
5 Predictive Coding and Information Bottleneck for Hallucination Detection in Large Language Models 提出基于预测编码和信息瓶颈的LLM幻觉检测框架，提升检测效率和可解释性 large language model
6 CogToM: A Comprehensive Theory of Mind Benchmark inspired by Human Cognition for Large Language Models CogToM：一个受人类认知启发的大语言模型心智理论综合基准 large language model
7 MapViT: A Two-Stage ViT-Based Framework for Real-Time Radio Quality Map Prediction in Dynamic Environments 提出MapViT，一种基于ViT的两阶段框架，用于动态环境中实时预测无线电质量地图。 large language model foundation model
8 EvoCUA: Evolving Computer Use Agents via Learning from Scalable Synthetic Experience EvoCUA：通过可扩展的合成经验学习进化计算机使用Agent foundation model multimodal
9 Sawtooth Wavefront Reordering: Enhanced CuTile FlashAttention on NVIDIA GB10 提出锯齿波前重排序技术，优化NVIDIA GB10上CuTile FlashAttention的L2缓存性能。 large language model
10 Connect the Dots: Knowledge Graph-Guided Crawler Attack on Retrieval-Augmented Generation Systems 提出RAGCRAWLER，利用知识图谱引导爬虫攻击检索增强生成系统，提升隐私窃取效率。 large language model
11 Data-Free Privacy-Preserving for LLMs via Model Inversion and Selective Unlearning 提出Data-Free Selective Unlearning (DFSU)，解决LLM在无训练数据下的隐私保护问题。 large language model
12 LLM Prompt Evaluation for Educational Applications 提出一种系统化LLM Prompt评估方法，用于提升教育应用中个性化教学效果。 large language model
13 Controlling Long-Horizon Behavior in Language Model Agents with Explicit State Dynamics 提出基于显式状态动态的语言模型Agent控制方法，提升长程交互行为一致性。 large language model
14 Deja Vu in Plots: Leveraging Cross-Session Evidence with Retrieval-Augmented LLMs for Live Streaming Risk Assessment 提出CS-VAR，利用检索增强LLM进行直播风险跨会话证据感知评估 large language model
15 Tabular Incremental Inference 提出Tabular Incremental Inference以解决动态表格推理问题 large language model
16 CoNRec: Context-Discerning Negative Recommendation with LLMs 提出CoNRec以解决用户负面偏好建模问题 large language model
17 Autonomous Business System via Neuro-symbolic AI AUTOBUS：基于神经符号AI的自主业务系统，解决企业流程重构难题。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (5 篇)

#题目一句话要点标签🔗⭐
18 From Passive Metric to Active Signal: The Evolving Role of Uncertainty Quantification in Large Language Models 大型语言模型中不确定性量化的演进：从被动指标到主动信号 reinforcement learning large language model
19 Decoupling Return-to-Go for Efficient Decision Transformer 提出解耦决策Transformer（DDT），提升离线强化学习效率与性能。 reinforcement learning offline RL offline reinforcement learning
20 Off-Policy Actor-Critic with Sigmoid-Bounded Entropy for Real-World Robot Learning 提出SigEnt-SAC算法，利用单条专家轨迹实现真实机器人环境中的高效强化学习。 reinforcement learning SAC VLA
21 Structured Hints for Sample-Efficient Lean Theorem Proving 利用结构化提示提升精简定理证明的样本效率 reinforcement learning large language model
22 PhysProver: Advancing Automatic Theorem Proving for Physics PhysProver：首个物理领域自动定理证明框架，提升物理及数学推理能力。 reinforcement learning foundation model

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
23 Cosmos Policy: Fine-Tuning Video Models for Visuomotor Control and Planning Cosmos Policy：通过微调视频模型实现视觉运动控制与规划 manipulation bi-manual bimanual manipulation

⬅️ 返回 cs.AI 首页 · 🏠 返回主页