cs.AI（2025-12-18）

📊 共 33 篇论文 | 🔗 5 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (27 🔗4) 支柱二：RL算法与架构 (RL & Architecture) (3) 支柱一：机器人控制 (Robot Control) (2 🔗1) 支柱三：空间感知与语义 (Perception & Semantics) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (27 篇)

#题目一句话要点标签🔗⭐
1 AlignMerge - Alignment-Preserving Large Language Model Merging via Fisher-Guided Geometric Constraints AlignMerge：通过Fisher引导的几何约束实现对齐保持的大语言模型合并 large language model foundation model instruction following
2 TOGGLE: Temporal Logic-Guided Large Language Model Compression for Edge TOGGLE：时序逻辑引导的大语言模型边缘压缩方法 large language model
3 Prefix Probing: Lightweight Harmful Content Detection for Large Language Models 提出Prefix Probing，以低延迟、低成本实现大语言模型有害内容检测。 large language model
4 TimeSeries2Report prompting enables adaptive large language model management of lithium-ion batteries 提出TimeSeries2Report框架，实现大语言模型对锂离子电池的自适应管理 large language model
5 A Multi-Agent Large Language Model Framework for Automated Qualitative Analysis 提出CoTI：一个基于多Agent LLM的自动化定性分析框架，应用于心力衰竭患者访谈分析。 large language model
6 Do Multi-Agents Solve Better Than Single? Evaluating Agentic Frameworks for Diagram-Grounded Geometry Problem Solving and Reasoning 对比单智能体与多智能体框架，评估其在图解几何问题求解中的性能差异 large language model multimodal ✅
7 Scaling Laws for Energy Efficiency of Local LLMs 针对本地LLM，揭示CPU能效缩放规律，并提出量子启发压缩优化方案。 large language model multimodal
8 Code-in-the-Loop Forensics: Agentic Tool Use for Image Forgery Detection 提出ForenAgent，利用Agentic工具解决图像伪造检测中跨层信息融合难题。 large language model multimodal
9 AMUSE: Audio-Visual Benchmark and Alignment Framework for Agentic Multi-Speaker Understanding AMUSE：用于Agentic多说话人理解的视听基准和对齐框架 large language model multimodal
10 Scaling Text2SQL via LLM-efficient Schema Filtering with Functional Dependency Graph Rerankers 提出GRaST，通过LLM高效的模式过滤和函数依赖图重排序，扩展Text2SQL系统处理大规模数据库的能力。 large language model ✅
11 Plausibility as Failure: How LLMs and Humans Co-Construct Epistemic Error 揭示LLM与人类交互中认知错误的共建机制，强调评估的解释性视角 large language model
12 Cyber Humanism in Education: Reclaiming Agency through AI and Learning Sciences 提出教育领域“赛博人文主义”框架，通过AI与学习科学重塑人类能动性 large language model
13 Microsoft Academic Graph Information Retrieval for Research Recommendation and Assistance 提出基于注意力的子图检索器，用于科研推荐和辅助，提升知识推理能力。 large language model
14 From Personalization to Prejudice: Bias and Discrimination in Memory-Enhanced AI Agents for Recruitment 揭示记忆增强型AI招聘Agent中的偏见引入与强化机制 large language model
15 cuPilot: A Strategy-Coordinated Multi-agent Framework for CUDA Kernel Evolution cuPilot：一种策略协调的多智能体框架，用于CUDA内核演化 large language model ✅
16 Towards AI-Supported Research: a Vision of the TIB AIssistant 提出TIB AIssistant：一个支持AI的跨学科科研协作平台 large language model
17 TIB AIssistant: a Platform for AI-Supported Research Across Research Life Cycles TIB AIssistant：一个支持研究全生命周期的人工智能研究平台 large language model
18 Introducing ORKG ASK: an AI-driven Scholarly Literature Search and Exploration System Taking a Neuro-Symbolic Approach 提出ORKG ASK：一种基于神经符号方法的AI驱动的学术文献搜索与探索系统 large language model
19 Synthelite: Chemist-aligned and feasibility-aware synthesis planning with LLMs Synthelite：利用LLM实现化学家友好且可行性感知的合成路线规划 large language model
20 AI Needs Physics More Than Physics Needs AI 强调物理学对人工智能发展的重要性，呼吁融合理论严谨性与机器学习灵活性。 large language model
21 Agent Tools Orchestration Leaks More: Dataset, Benchmark, and Mitigation 揭示Agent工具编排中的隐私泄露风险，并提出TOP-Bench基准与PEP缓解方法 large language model
22 Beyond the Benchmark: Innovative Defenses Against Prompt Injection Attacks 针对LLaMA模型，提出迭代式防御框架，提升抵御Prompt注入攻击能力 chain-of-thought
23 Adaptation of Agentic AI 构建Agentic AI自适应框架，提升智能体性能、可靠性和泛化能力 foundation model
24 QuadSentinel: Sequent Safety for Machine-Checkable Control in Multi-agent Systems QuadSentinel：多智能体系统中基于时序逻辑的安全可验证控制 large language model ✅
25 Beyond Blind Spots: Analytic Hints for Mitigating LLM-Based Evaluation Pitfalls 利用分析提示缓解LLM代码评估中的盲点，提升COBOL代码生成质量 large language model
26 Learning to Wait: Synchronizing Agents with the Physical World 提出Agent侧时间同步方法，解决LLM在异步环境中的时序认知问题 large language model
27 Ev-Trust: A Strategy Equilibrium Trust Mechanism for Evolutionary Games in LLM-Based Multi-Agent Services 提出Ev-Trust机制，利用演化博弈论解决LLM多智能体服务中的信任问题。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (3 篇)

#题目一句话要点标签🔗⭐
28 Generative Adversarial Reasoner: Enhancing LLM Reasoning with Adversarial Reinforcement Learning 提出Generative Adversarial Reasoner，通过对抗强化学习提升LLM的推理能力。 reinforcement learning distillation reward shaping
29 The Evolution of Reranking Models in Information Retrieval: From Heuristic Methods to Large Language Models 综述信息检索中重排序模型演进：从启发式方法到大型语言模型 distillation large language model
30 Coordinated Anti-Jamming Resilience in Swarm Networks via Multi-Agent Reinforcement Learning 提出基于QMIX的多智能体强化学习方法，提升集群网络抗干扰能力 reinforcement learning

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
31 Hypernetworks That Evolve Themselves 提出自引用图超网络，实现无需外部优化器的神经网络自进化。 locomotion reinforcement learning
32 OS-Oracle: A Comprehensive Framework for Cross-Platform GUI Critic Models OS-Oracle：跨平台GUI评论模型的综合框架，提升计算机使用代理的决策能力 manipulation ✅

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

#题目一句话要点标签🔗⭐
33 Distributional AGI Safety 提出分布式AGI安全框架，通过虚拟沙盒经济保障群体智能安全。 affordance

⬅️ 返回 cs.AI 首页 · 🏠 返回主页