cs.AI（2025-10-27）

📊 共 35 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) (21 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (9) 支柱一：机器人控制 (Robot Control) (4) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (21 篇)

#题目一句话要点标签🔗⭐
1 Game-TARS: Pretrained Foundation Models for Scalable Generalist Multimodal Game Agents Game-TARS：基于预训练通用多模态游戏Agent，实现跨域可扩展性 generalist agent foundation model multimodal
2 Emotion-Coherent Reasoning for Multimodal LLMs via Emotional Rationale Verifier 提出情感推理验证器，提升多模态LLM情感理解与解释一致性 large language model multimodal
3 Beyond the Failures: Rethinking Foundation Models in Pathology 病理学领域需重新思考基础模型，避免盲目套用自然图像方法 foundation model
4 From Benchmarks to Business Impact: Deploying IBM Generalist Agent in Enterprise Production IBM提出CUGA通用Agent，并应用于企业BPO人才招聘领域，验证其业务价值。 generalist agent ✅
5 Decentralized Multi-Agent Goal Assignment for Path Planning using Large Language Models 利用大语言模型实现去中心化多智能体目标分配与路径规划 large language model
6 ReCAP: Recursive Context-Aware Reasoning and Planning for Large Language Model Agents ReCAP：用于大语言模型Agent的递归上下文感知推理与规划 large language model
7 Alita-G: Self-Evolving Generative Agent for Agent Generation ALITA-G：一种自进化生成式Agent，通过生成、抽象和管理MCP工具实现Agent的领域专家化。 generalist agent large language model
8 JanusCoder: Towards a Foundational Visual-Programmatic Interface for Code Intelligence JanusCoder：构建用于代码智能的基础视觉-程序化接口 multimodal ✅
9 PRO: Enabling Precise and Robust Text Watermark for Open-Source LLMs PRO：为开源LLM实现精确且鲁棒的文本水印，提升知识产权保护能力 large language model
10 ReCode: Unify Plan and Action for Universal Granularity Control ReCode：通过统一计划与行动实现通用粒度控制 large language model ✅
11 AutoStreamPipe: LLM Assisted Automatic Generation of Data Stream Processing Pipelines AutoStreamPipe：利用LLM自动生成数据流处理管道，显著降低开发时间和错误率。 large language model
12 Smaller Models, Smarter Rewards: A Two-Sided Approach to Process and Outcome Rewards 利用小型语言模型，结合过程与结果奖励，提升代码生成质量 large language model
13 Evaluating the effectiveness of LLM-based interoperability 评估基于LLM的互操作性有效性，实现系统自主运行时互操作。 large language model
14 Agentic AI Security: Threats, Defenses, Evaluation, and Open Challenges 针对Agentic AI安全威胁，提出防御、评估框架及开放挑战分析 large language model
15 Mutual Wanting in Human--AI Interaction: Empirical Evidence from Large-Scale Analysis of GPT Model Transitions 提出互需对齐框架M-WAF，用于量化人机交互中用户期望与AI行为的匹配程度。 large language model
16 A Survey of Data Agents: Emerging Paradigm or Overstated Hype? 构建数据Agent分级体系，厘清自主程度，促进数据+AI生态发展。 large language model
17 Policy-Aware Generative AI for Safe, Auditable Data Access Governance 提出一种策略感知的生成式AI，用于安全、可审计的数据访问治理。 large language model
18 Exploring Vulnerability in AI Industry 提出AI脆弱性指数AIVI，量化评估AI产业上游价值链的系统性风险 foundation model
19 QueryIPI: Query-agnostic Indirect Prompt Injection on Coding Agents 提出QueryIPI，实现对编码Agent的查询无关型间接提示注入攻击 instruction following
20 RefleXGen:The unexamined code is not worth using RefleXGen：通过自反思提升LLM代码生成的安全性 large language model
21 Quantifying Document Impact in RAG-LLMs 提出Influence Score (IS)以量化RAG中单个文档对生成结果的影响。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (9 篇)

#题目一句话要点标签🔗⭐
22 USF-MAE: Ultrasound Self-Supervised Foundation Model with Masked Autoencoding USF-MAE：基于掩码自编码器的超声自监督预训练模型 MAE foundation model
23 Latent Chain-of-Thought for Visual Reasoning 提出基于隐式思维链的视觉推理方法，提升LVLM的泛化性和可解释性 reinforcement learning PPO chain-of-thought
24 Bid2X: Revealing Dynamics of Bidding Environment in Online Advertising from A Foundation Model Lens Bid2X：通过基础模型揭示在线广告竞价环境的动态特性 representation learning foundation model
25 Guiding Skill Discovery with Foundation Models 提出FoG方法，利用Foundation模型引导技能发现，避免不期望行为。 reinforcement learning foundation model
26 Human-Like Goalkeeping in a Realistic Football Simulation: a Sample-Efficient Reinforcement Learning Approach 提出一种高效强化学习方法，用于在真实足球游戏中训练类人守门员AI。 reinforcement learning deep reinforcement learning DRL
27 Multi-Agent Evolve: LLM Self-Improve through Co-evolution 提出Multi-Agent Evolve框架，通过多智能体协同进化提升LLM推理能力。 reinforcement learning MAE large language model
28 Think before Recommendation: Autonomous Reasoning-enhanced Recommender RecZero：提出基于强化学习的自主推理增强推荐系统，摆脱传统蒸馏范式。 reinforcement learning distillation large language model
29 CNOT Minimal Circuit Synthesis: A Reinforcement Learning Approach 提出基于强化学习的CNOT门最小化电路合成方法，超越现有技术。 reinforcement learning
30 Causal Deep Q Network 提出因果深度Q网络(Causal DQN)以提升强化学习中DQN的因果推理能力 reinforcement learning deep reinforcement learning

🔬 支柱一：机器人控制 (Robot Control) (4 篇)

#题目一句话要点标签🔗⭐
31 Hybrid Modeling, Sim-to-Real Reinforcement Learning, and Large Language Model Driven Control for Digital Twins 针对数字孪生，提出混合建模、Sim-to-Real强化学习及LLM驱动控制方法 sim-to-real MPC model predictive control
32 MFiSP: A Multimodal Fire Spread Prediction Framework 提出MFiSP多模态火灾蔓延预测框架，融合社交媒体与遥感数据提升预测精度。 manipulation multimodal
33 Learning Linearity in Audio Consistency Autoencoders via Implicit Regularization 通过隐式正则化学习音频一致性自编码器的线性特性 manipulation
34 Magentic Marketplace: An Open-Source Environment for Studying Agentic Markets 提出Magentic-Marketplace，用于研究LLM Agent在双边市场中的行为与动态。 manipulation

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

#题目一句话要点标签🔗⭐
35 Efficient and Encrypted Inference using Binarized Neural Networks within In-Memory Computing Architectures 提出一种基于PUF密钥的BNN加密推理方法，用于内存计算架构的安全加速。 OMOMO

⬅️ 返回 cs.AI 首页 · 🏠 返回主页