cs.AI（2025-12-29）

📊 共 25 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) (16 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (6 🔗3) 支柱一：机器人控制 (Robot Control) (1) 支柱三：空间感知与语义 (Perception & Semantics) (1) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (16 篇)

#题目一句话要点标签🔗⭐
1 Agentic Physical AI toward a Domain-Specific Foundation Model for Nuclear Reactor Control 提出Agentic Physical AI，用于核反应堆控制的领域特定基础模型。 foundation model multimodal
2 Breaking Audio Large Language Models by Attacking Only the Encoder: A Universal Targeted Latent-Space Audio Attack 提出一种通用目标潜在空间音频攻击，打破音频大语言模型编码器。 large language model multimodal
3 Toward Trustworthy Agentic AI: A Multimodal Framework for Preventing Prompt Injection Attacks 提出跨Agent多模态溯源防御框架，防范Agentic AI中的提示注入攻击 large language model multimodal
4 EquaCode: A Multi-Strategy Jailbreak Approach for Large Language Models via Equation Solving and Code Completion 提出EquaCode，利用方程求解与代码补全实现大语言模型的越狱攻击 large language model
5 How Large Language Models Systematically Misrepresent American Climate Opinions 揭示大型语言模型在美国气候观点上的系统性偏差，尤其是在交叉身份群体中。 large language model
6 Divergent-Convergent Thinking in Large Language Models for Creative Problem Generation CreativeDC：利用大语言模型中的发散-收敛思维生成多样化创意问题 large language model
7 SPIRAL: Symbolic LLM Planning via Grounded and Reflective Search SPIRAL：通过具身和反思搜索实现符号LLM规划 large language model chain-of-thought
8 From Model Choice to Model Belief: Establishing a New Measure for LLM-Based Research 提出“模型置信度”以更高效利用LLM的概率信息，提升模拟研究效率。 large language model
9 Enhancing Temporal Awareness in LLMs for Temporal Point Processes 提出TPP-TAL框架，增强LLM在时序点过程中的时间感知能力 large language model ✅
10 It's a TRAP! Task-Redirecting Agent Persuasion Benchmark for Web Agents 提出TRAP基准以评估网络代理的劝说脆弱性 large language model
11 AKG kernel Agent: A Multi-Agent Framework for Cross-Platform Kernel Synthesis 提出AKG内核代理以解决跨平台内核合成问题 multimodal
12 CASCADE: Cumulative Agentic Skill Creation through Autonomous Development and Evolution CASCADE：通过自主开发和演化实现累积式智能体技能创造 large language model
13 From Correctness to Collaboration: Toward a Human-Centered Framework for Evaluating AI Agent Behavior in Software Engineering 提出人本框架以评估软件工程中AI代理行为 large language model
14 The Gaining Paths to Investment Success: Information-Driven LLM Graph Reasoning for Venture Capital Prediction 提出MIRAGE-VC，利用信息增益驱动的LLM图推理进行风险投资预测。 large language model
15 Securing the AI Supply Chain: What Can We Learn From Developer-Reported Security Issues and Solutions of AI Projects? 分析AI项目开发者报告的安全问题与解决方案，保障AI供应链安全。 large language model
16 TCEval: Using Thermal Comfort to Assess Cognitive and Perceptual Abilities of AI TCEval：利用热舒适度评估AI的认知和感知能力 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (6 篇)

#题目一句话要点标签🔗⭐
17 Replay Failures as Successes: Sample-Efficient Reinforcement Learning for Instruction Following 提出HiR：一种面向指令跟随任务的、基于回溯重放的样本高效强化学习方法 reinforcement learning preference learning large language model ✅
18 InSPO: Unlocking Intrinsic Self-Reflection for LLM Preference Optimization InSPO：通过内省自反优化提升LLM偏好对齐 RLHF DPO direct preference optimization
19 Web World Models 提出Web World Model，结合Web代码的可靠性与LLM的生成能力，构建可控且开放的Agent环境。 world model large language model ✅
20 Alpha-R1: Alpha Screening with LLM Reasoning via Reinforcement Learning 提出Alpha-R1，利用强化学习训练LLM进行上下文感知的Alpha筛选，提升投资策略的鲁棒性。 reinforcement learning large language model ✅
21 Prompt-Induced Over-Generation as Denial-of-Service: A Black-Box Attack-Side Benchmark 提出黑盒攻击基准，研究提示诱导的大语言模型过度生成漏洞，可用于拒绝服务攻击。 reinforcement learning large language model
22 AI-Native Integrated Sensing and Communications for Self-Organizing Wireless Networks: Architectures, Learning Paradigms, and System-Level Design 提出AI原生集成感知通信框架，赋能自组织无线网络资源优化。 reinforcement learning deep reinforcement learning

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
23 MindWatcher: Toward Smarter Multimodal Tool-Integrated Reasoning 提出MindWatcher，一种集成多模态工具的智能推理Agent，用于解决复杂决策任务。 manipulation multimodal chain-of-thought

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

#题目一句话要点标签🔗⭐
24 An Inference-Based Architecture for Intent and Affordance Saturation in Decision-Making 提出基于推理的架构，解决决策中意图和可供性饱和导致的决策瘫痪问题 affordance

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
25 Autoregressive long-horizon prediction of plasma edge dynamics 提出基于Transformer的自回归模型，用于高效预测等离子体边缘动力学 spatiotemporal

⬅️ 返回 cs.AI 首页 · 🏠 返回主页