cs.AI（2026-03-03）

📊 共 33 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) (23 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (7 🔗1) 支柱一：机器人控制 (Robot Control) (2 🔗1) 支柱六：视频提取与匹配 (Video Extraction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (23 篇)

#题目一句话要点标签🔗⭐
1 See and Remember: A Multimodal Agent for Web Traversal 提出V-GEMS，解决LLM智能体Web导航中的空间迷失和循环问题 large language model multimodal visual grounding ✅
2 ShipTraj-R1: Reinforcing Ship Trajectory Prediction in Large Language Models via Group Relative Policy Optimization 提出ShipTraj-R1，利用大语言模型和强化学习优化船舶轨迹预测。 large language model chain-of-thought
3 LLM-MLFFN: Multi-Level Autonomous Driving Behavior Feature Fusion via Large Language Model 提出LLM-MLFFN，利用大语言模型融合多层次特征，提升自动驾驶行为分类精度。 large language model
4 Detecting Structural Heart Disease from Electrocardiograms via a Generalized Additive Model of Interpretable Foundation-Model Predictors 提出基于可解释 ECG 基础模型预测器的广义加性模型，用于心血管疾病检测。 foundation model
5 NeuroProlog: Multi-Task Fine-Tuning for Neurosymbolic Mathematical Reasoning via the Cocktail Effect 提出NeuroProlog以解决数学推理中的逻辑不一致问题 large language model symbolic grounding
6 SorryDB: Can AI Provers Complete Real-World Lean Theorems? 提出SorryDB：一个动态更新的Lean定理证明基准，用于评估AI证明器的能力。 large language model
7 AI-for-Science Low-code Platform with Bayesian Adversarial Multi-Agent Framework 提出基于贝叶斯对抗多智能体框架的AI for Science低代码平台，提升科学代码生成可靠性。 large language model
8 Type-Aware Retrieval-Augmented Generation with Dependency Closure for Solver-Executable Industrial Optimization Modeling 提出类型感知检索增强生成方法，解决工业优化建模中模型可执行性问题。 large language model
9 Saarthi for AGI: Towards Domain-Specific General Intelligence for Formal Verification Saarthi框架通过规则和RAG增强，提升形式验证领域特定通用智能。 large language model
10 Agentic AI-based Coverage Closure for Formal Verification 提出基于代理AI的覆盖闭合方法以提升形式验证效率 large language model
11 Beyond Task Completion: Revealing Corrupt Success in LLM Agents through Procedure-Aware Evaluation 提出程序感知评估（PAE）框架，揭示LLM Agent任务完成中的隐蔽性错误。 large language model
12 REGAL: A Registry-Driven Architecture for Deterministic Grounding of Agentic AI in Enterprise Telemetry REGAL：一种注册表驱动架构，用于企业遥测中Agentic AI的确定性基础 large language model
13 OrchMAS: Orchestrated Reasoning with Multi Collaborative Heterogeneous Scientific Expert Structured Agents OrchMAS：提出多智能体协同框架，解决科学领域复杂推理难题 large language model
14 Architecting Trust in Artificial Epistemic Agents 构建可信赖的认知AI Agent，应对知识生态系统中的挑战。 large language model
15 SEALing the Gap: A Reference Framework for LLM Inference Carbon Estimation via Multi-Benchmark Driven Embodiment 提出LLM推理碳排放估算框架以应对可持续性挑战 large language model
16 LLM-based Argument Mining meets Argumentation and Description Logics: a Unified Framework for Reasoning about Debates 提出融合论证挖掘、论证逻辑与描述逻辑的统一框架，用于辩论推理。 large language model
17 Agentified Assessment of Logical Reasoning Agents 提出基于Agent的逻辑推理评估框架，提升评估的可复现性、可审计性和鲁棒性。 chain-of-thought
18 Rethinking Code Similarity for Automated Algorithm Design with LLMs 提出BehaveSim，通过行为相似性度量提升LLM驱动的算法自动设计。 large language model ✅
19 EvoSkill: Automated Skill Discovery for Multi-Agent Systems 提出EvoSkill以自动发现多智能体系统中的技能 zero-shot transfer
20 A Natural Language Agentic Approach to Study Affective Polarization 提出基于自然语言Agent的框架，用于研究社交媒体中的情感极化现象 large language model
21 LiveAgentBench: Comprehensive Benchmarking of Agentic Systems Across 104 Real-World Challenges LiveAgentBench：包含104个真实世界挑战的Agentic系统综合基准测试 large language model
22 A Neuropsychologically Grounded Evaluation of LLM Cognitive Abilities 提出NeuroCognition基准，从神经心理学角度评估LLM认知能力 large language model
23 Human-Certified Module Repositories for the AI Age 提出人工认证模块仓库HCMRs，保障AI辅助开发时代软件可信度 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (7 篇)

#题目一句话要点标签🔗⭐
24 LLMs for High-Frequency Decision-Making: Normalized Action Reward-Guided Consistency Policy Optimization 提出NAR-CP方法，解决LLM在高频决策任务中的策略失准问题 consistency policy reward shaping large language model
25 TikZilla: Scaling Text-to-TikZ with High-Quality Data and Reinforcement Learning TikZilla：通过高质量数据和强化学习扩展文本到TikZ的生成能力 reinforcement learning large language model
26 Enhancing User Throughput in Multi-panel mmWave Radio Access Networks for Beam-based MU-MIMO Using a DRL Method 提出基于DRL的波束管理方法，提升毫米波MU-MIMO系统用户吞吐量。 reinforcement learning deep reinforcement learning DRL
27 Retrievit: In-context Retrieval Capabilities of Transformers, State Space Models, and Hybrid Architectures 研究混合架构Transformer与SSM在上下文检索中的能力，探索其在数据效率和泛化性上的优势。 SSM state space model
28 RAPO: Expanding Exploration for LLM Agents via Retrieval-Augmented Policy Optimization RAPO：通过检索增强策略优化扩展LLM Agent的探索能力 reinforcement learning large language model
29 SAE as a Crystal Ball: Interpretable Features Predict Cross-domain Transferability of LLMs without Training 提出基于稀疏自编码器的迁移性评分（STS），无需训练即可预测LLM跨域迁移能力。 reinforcement learning large language model ✅
30 QFlowNet: Fast, Diverse, and Efficient Unitary Synthesis with Generative Flow Networks QFlowNet：利用生成流网络实现快速、多样且高效的酉矩阵合成 reinforcement learning reward shaping

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
31 Design Generative AI for Practitioners: Exploring Interaction Approaches Aligned with Creative Practice 为设计从业者设计生成式AI：探索与创造性实践对齐的交互方法 manipulation
32 Credibility Governance: A Social Mechanism for Collective Self-Correction under Weak Truth Signals 提出可信度治理机制以解决在线平台集体判断脆弱性问题 manipulation ✅

🔬 支柱六：视频提取与匹配 (Video Extraction) (1 篇)

#题目一句话要点标签🔗⭐
33 SpatialText: A Pure-Text Cognitive Benchmark for Spatial Understanding in Large Language Models SpatialText：用于评估大语言模型空间理解能力的纯文本认知基准 egocentric large language model multimodal

⬅️ 返回 cs.AI 首页 · 🏠 返回主页