cs.AI（2026-04-30）

📊 共 50 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) (36 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (9) 支柱一：机器人控制 (Robot Control) (2) 支柱三：空间感知与语义 (Perception & Semantics) (2) 支柱七：动作重定向 (Motion Retargeting) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (36 篇)

#题目一句话要点标签🔗⭐
1 From Mirage to Grounding: Towards Reliable Multimodal Circuit-to-Verilog Code Generation 揭示MLLM电路图到Verilog代码生成中的“幻影”现象，提出VeriGround模型提升可靠性。 large language model multimodal visual grounding
2 Heterogeneous Scientific Foundation Model Collaboration Eywa：异构科学基础模型协作框架，扩展Agentic LLM在科学领域的应用 large language model foundation model
3 InteractWeb-Bench: Can Multimodal Agent Escape Blind Execution in Interactive Website Generation? InteractWeb-Bench：评估多模态Agent在交互式网站生成中避免盲目执行的能力 large language model multimodal
4 Design Structure Matrix Modularization with Large Language Models 利用大语言模型进行设计结构矩阵模块化，无需专业优化代码。 large language model
5 Iterative Multimodal Retrieval-Augmented Generation for Medical Question Answering 提出MED-VRAG，一种迭代多模态检索增强生成框架，用于医学问答。 multimodal
6 Belief-Guided Inference Control for Large Language Model Services via Verifiable Observations 提出Veroic框架，通过可验证观测实现大语言模型服务中风险感知的推理控制。 large language model
7 SpecVQA: A Benchmark for Spectral Understanding and Visual Question Answering in Scientific Images 提出SpecVQA：科学图像中光谱理解与视觉问答的专业评测基准。 large language model multimodal
8 The Effects of Visual Priming on Cooperative Behavior in Vision-Language Models 研究视觉启动对视觉语言模型合作行为的影响，以迭代囚徒困境为测试场景 chain-of-thought
9 Exploring Interaction Paradigms for LLM Agents in Scientific Visualization 探索LLM Agent在科学可视化中的交互范式，权衡性能、鲁棒性和灵活性。 large language model
10 Math Education Digital Shadows for facilitating learning with LLMs: Math performance, anxiety and confidence in simulated students and AIs 提出MEDS数据集，用于评估LLM在数学教育中的能力、偏差及心理特征。 large language model
11 What Makes a Good Terminal-Agent Benchmark Task: A Guideline for Adversarial, Difficult, and Legible Evaluation Design 为终端代理基准测试任务设计提供指导，强调对抗性、难度和可读性 large language model
12 Collaborative Agent Reasoning Engineering (CARE): A Three-Party Design Methodology for Systematically Engineering AI Agents with Subject Matter Experts, Developers, and Helper Agents CARE：一种三方协作的AI Agent工程方法，提升科学领域LLM Agent开发效率 large language model
13 LLMs as ASP Programmers: Self-Correction Enables Task-Agnostic Nonmonotonic Reasoning 提出LLM+ASP框架，利用自校正实现任务无关的非单调推理 large language model
14 MM-StanceDet: Retrieval-Augmented Multi-modal Multi-agent Stance Detection 提出MM-StanceDet，通过检索增强的多智能体框架解决多模态立场检测中的融合难题。 multimodal
15 MCPHunt: An Evaluation Framework for Cross-Boundary Data Propagation in Multi-Server MCP Agents MCPHunt：多服务器MCP代理中跨边界数据传播的评估框架 instruction following
16 Post-Optimization Adaptive Rank Allocation for LoRA 提出PARA，一种LoRA后优化自适应秩分配方法，提升参数效率。 foundation model
17 Test Before You Deploy: Governing Updates in the LLM Supply Chain 提出LLM供应链治理框架，保障部署端LLM更新的兼容性和安全性 large language model
18 RuC: HDL-Agnostic Rule Completion Benchmark Generation RuC：一种与硬件描述语言无关的、基于规则的可控代码补全基准生成框架 large language model
19 Intent2Tx: Benchmarking LLMs for Translating Natural Language Intents into Ethereum Transactions Intent2Tx：构建基准测试，评估LLM将自然语言意图转化为以太坊交易的能力 large language model
20 Position-Aware Drafting for Inference Acceleration in LLM-Based Generative List-Wise Recommendation PAD-Rec：针对LLM生成式列表推荐的位置感知草稿加速推理 large language model
21 AgentEconomist: An End-to-end Agentic System Translating Economic Intuitions into Executable Computational Experiments AgentEconomist：将经济学直觉转化为可执行计算实验的端到端智能系统 large language model
22 Bridging Values and Behavior: A Hierarchical Framework for Proactive Embodied Agents 提出ValuePlanner，解决具身智能体长期自主行为决策问题 instruction following
23 When Agents Evolve, Institutions Follow 提出基于历史政治制度的多智能体架构，提升LLM的集体智能。 large language model ✅
24 HAVEN: Hybrid Automated Verification ENgine for UVM Testbench Synthesis with LLMs HAVEN：一种混合自动化验证引擎，利用LLM进行UVM测试平台综合 large language model
25 Optimization before Evaluation: Evaluation with Unoptimised Prompts Can be Misleading 揭示LLM评估陷阱：未优化Prompt可能导致模型排序失真 large language model
26 Political Bias Audits of LLMs Capture Sycophancy to the Inferred Auditor 揭示大型语言模型的政治偏见与迎合审计者的关系 large language model
27 In-Context Examples Suppress Scientific Knowledge Recall in LLMs 在LLM中，上下文示例会抑制科学知识的调用，导致模型倾向于经验模式拟合。 large language model
28 Security Attack and Defense Strategies for Autonomous Agent Frameworks: A Layered Review with OpenClaw as a Case Study 针对自主Agent框架的安全风险，提出分层分析与防御策略综述 large language model
29 Toward Autonomous SOC Operations: End-to-End LLM Framework for Threat Detection, Query Generation, and Resolution in Security Operations 提出端到端LLM框架，自动化SOC威胁检测、查询生成和事件响应。 large language model
30 METASYMBO: Multi-Agent Language-Guided Metamaterial Discovery via Symbolic Latent Evolution 提出MetaSymbO，通过符号驱动的潜在演化实现语言引导的多智能体超材料发现。 large language model
31 Evaluating Epistemic Guardrails in AI Reading Assistants: A Behavioral Audit of a Minimal Prototype 提出评估AI阅读助手认知防护栏的协议，揭示交互行为动态及边界功能。 large language model
32 ARMOR 2025: A Military-Aligned Benchmark for Evaluating Large Language Model Safety Beyond Civilian Contexts 提出ARMOR 2025：一个面向军事场景的大语言模型安全评估基准 large language model
33 Minimal, Local, Causal Explanations for Jailbreak Success in Large Language Models 提出LOCA方法，为大语言模型越狱攻击提供最小化、局部化和因果解释 large language model
34 DeGenTWeb: A First Look at LLM-dominant Websites DeGenTWeb：首次系统性识别并分析LLM主导的网站，揭示其普遍性和演变趋势 large language model
35 A Survey of Reasoning-Intensive Retrieval: Progress and Challenges 针对推理密集型检索的综述：系统性地分析现有方法，并展望未来研究方向 large language model
36 TADI: Tool-Augmented Drilling Intelligence via Agentic LLM Orchestration over Heterogeneous Wellsite Data TADI：通过Agentic LLM编排异构井场数据，实现工具增强的钻井智能 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🔗⭐
37 Rethinking Agentic Reinforcement Learning In Large Language Models 基于大语言模型的Agentic强化学习：重新思考智能体自主性 reinforcement learning large language model
38 WaferSAGE: Large Language Model-Powered Wafer Defect Analysis via Synthetic Data Generation and Rubric-Guided Reinforcement Learning WaferSAGE：利用合成数据和规则引导强化学习进行晶圆缺陷分析 reinforcement learning large language model
39 Simulating clinical interventions with a generative multimodal model of human physiology HealthFormer：基于Transformer的生成式多模态模型，用于模拟临床干预和预测生理轨迹。 world model world models multimodal
40 GUI Agents with Reinforcement Learning: Toward Digital Inhabitants 提出GUI智能体研究综述，探索强化学习在GUI自动化中的应用及未来方向 reinforcement learning offline RL world model
41 Graph World Models: Concepts, Taxonomy, and Future Directions 提出图世界模型（GWM）概念，系统性地解决传统世界模型在复杂环境中的局限性。 world model world models
42 LLM as Clinical Graph Structure Refiner: Enhancing Representation Learning in EEG Seizure Diagnosis 提出基于LLM的临床图结构精炼方法，提升脑电癫痫诊断的表征学习 representation learning large language model
43 Towards Neuro-symbolic Causal Rule Synthesis, Verification, and Evaluation Grounded in Legal and Safety Principles 提出神经符号因果规则合成框架，用于安全关键场景下的规则自动生成与验证。 reinforcement learning deep reinforcement learning large language model
44 TUR-DPO: Topology- and Uncertainty-Aware Direct Preference Optimization 提出TUR-DPO，一种拓扑和不确定性感知的直接偏好优化方法，提升LLM推理能力。 reinforcement learning PPO RLHF
45 XekRung Technical Report XekRung：面向网络安全的先进大规模语言模型 reinforcement learning large language model

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
46 PRTS: A Primitive Reasoning and Tasking System via Contrastive Representations PRTS：通过对比表示进行原始推理和任务处理的机器人VLA基础模型 manipulation reachability-aware reinforcement learning
47 RHyVE: Competence-Aware Verification and Phase-Aware Deployment for LLM-Generated Reward Hypotheses RHyVE：针对LLM生成奖励假设，提出能力感知验证和阶段感知部署方法 manipulation reinforcement learning reward design

🔬 支柱三：空间感知与语义 (Perception & Semantics) (2 篇)

#题目一句话要点标签🔗⭐
48 A Pattern Language for Resilient Visual Agents 提出一种视觉Agent架构模式语言，解决企业环境中视觉语言行为模型的集成挑战。 affordance vision-language-action VLA
49 Knowledge Affordances for Hybrid Human-AI Information Seeking 提出知识可供性(KA)概念，用于混合人机环境中智能体的信息寻求决策。 affordance

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

#题目一句话要点标签🔗⭐
50 SpatialGrammar: A Domain-Specific Language for LLM-Based 3D Indoor Scene Generation 提出SpatialGrammar领域特定语言，提升LLM生成3D室内场景的空间一致性。 spatial relationship embodied AI

⬅️ 返回 cs.AI 首页 · 🏠 返回主页