cs.AI（2026-04-09）

📊 共 48 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) (30 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (15 🔗3) 支柱六：视频提取与匹配 (Video Extraction) (1) 支柱一：机器人控制 (Robot Control) (1) 支柱三：空间感知与语义 (Perception & Semantics) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (30 篇)

#题目一句话要点标签🔗⭐
1 How Far Are Large Multimodal Models from Human-Level Spatial Action? A Benchmark for Goal-Oriented Embodied Navigation in Urban Airspace 构建城市空域导航基准，评估大型多模态模型在具身空间行为中的能力 vision-language-action multimodal ✅
2 MONETA: Multimodal Industry Classification through Geographic Information with Multi Agent Systems MONETA：利用地理信息和多智能体系统进行多模态行业分类 large language model multimodal
3 CIAO - Code In Architecture Out - Automated Software Architecture Documentation with Large Language Models CIAO：利用大语言模型自动生成软件架构文档，提升系统可理解性。 large language model
4 ImplicitMemBench: Measuring Unconscious Behavioral Adaptation in Large Language Models 提出ImplicitMemBench，用于评估大语言模型中无意识行为适应能力的基准测试。 large language model
5 MIMIC-Py: An Extensible Tool for Personality-Driven Automated Game Testing with Large Language Models MIMIC-Py：基于LLM的性格驱动型自动化游戏测试可扩展工具 large language model ✅
6 Emotion Concepts and their Function in a Large Language Model 发现大语言模型中功能性情绪：情绪概念影响模型行为与对齐 large language model
7 Learning Who Disagrees: Demographic Importance Weighting for Modeling Annotator Distributions with DiADEM 提出DiADEM模型，通过人口统计学重要性加权建模标注者分布，提升主观内容理解。 large language model chain-of-thought
8 Wiring the 'Why': A Unified Taxonomy and Survey of Abductive Reasoning in LLMs 构建演绎推理统一分类法，并对LLM中的溯因推理进行全面调研。 large language model
9 Are we still able to recognize pearls? Machine-driven peer review and the risk to creativity: An explainable RAG-XAI detection framework with markers extraction 提出RAG-XAI框架，用于检测机器驱动的同行评审，保障科研创造力。 large language model
10 Visual Perceptual to Conceptual First-Order Rule Learning Networks 提出γILP框架，解决从图像数据中学习一阶规则并自动生成谓词的难题。 large language model
11 From Safety Risk to Design Principle: Peer-Preservation in Multi-Agent LLM Systems and Its Implications for Orchestrated Democratic Discourse Analysis 针对多Agent LLM系统中同伴保护现象，提出基于身份匿名化的民主讨论分析架构设计 large language model
12 Verify Before You Commit: Towards Faithful Reasoning in LLM Agents via Self-Auditing 提出SAVeR框架，通过自审计保证LLM Agent推理过程的忠实性。 large language model
13 SkillClaw: Let Skills Evolve Collectively with Agentic Evolver SkillClaw：通过Agentic Evolver实现技能的集体进化，提升多用户Agent生态系统性能 large language model
14 Don't Overthink It: Inter-Rollout Action Agreement as a Free Adaptive-Compute Signal for LLM Agents TrACE：基于行动一致性的LLM Agent自适应计算控制器 large language model
15 Neural-Symbolic Knowledge Tracing: Injecting Educational Knowledge into Deep Learning for Responsible Learner Modelling 提出Responsible-DKT，融合教育知识的神经符号知识追踪方法，提升学习者建模的责任性。 large language model
16 IoT-Brain: Grounding LLMs for Semantic-Spatial Sensor Scheduling IoT-Brain：通过空间轨迹图STG连接LLM与物理世界，实现语义空间传感器调度 large language model
17 DialBGM: A Benchmark for Background Music Recommendation from Everyday Multi-Turn Dialogues DialBGM：提出一个日常多轮对话背景音乐推荐的基准数据集。 multimodal
18 An Agentic Evaluation Architecture for Historical Bias Detection in Educational Textbooks 提出Agentic评估架构，用于检测教育教科书中存在的历史偏见。 multimodal
19 PyVRP$^+$: LLM-Driven Metacognitive Heuristic Evolution for Hybrid Genetic Search in Vehicle Routing Problems PyVRP$^+$：基于LLM驱动的元认知启发式进化，用于车辆路径问题中的混合遗传搜索 large language model
20 SPARD: Self-Paced Curriculum for RL Alignment via Integrating Reward Dynamics and Data Utility SPARD：通过整合奖励动态和数据效用，实现强化学习对齐的自步课程学习 large language model
21 Filling the Gaps: Selective Knowledge Augmentation for LLM Recommenders 提出KnowSA_CKP，通过选择性知识增强提升LLM推荐器的性能和效率 large language model
22 More Capable, Less Cooperative? When LLMs Fail At Zero-Cost Collaboration 揭示LLM在零成本协作中失效的原因，强调智能扩展并非解决多智能体协作的唯一途径 large language model
23 Squeeze Evolve: Unified Multi-Model Orchestration for Verifier-Free Evolution Squeeze Evolve：用于无验证器进化的统一多模型编排框架 multimodal
24 Towards Knowledgeable Deep Research: Framework and Benchmark 提出混合知识分析框架HKA，解决深度研究中结构化与非结构化知识融合问题。 multimodal
25 Multi-Agent Orchestration for High-Throughput Materials Screening on a Leadership-Class System 提出基于多Agent协同的高通量材料筛选框架，提升HPC系统利用率。 large language model
26 MONETA: Multimodal Industry Classification through Geographic Information with Multi Agent Systems MONETA：利用地理信息和多智能体系统进行多模态行业分类 large language model multimodal
27 Demystifying the Silence of Correctness Bugs in PyTorch Compiler 针对PyTorch编译器正确性Bug，提出基于LLM变异的检测方法AlignGuard large language model
28 Model Space Reasoning as Search in Feedback Space for Planning Domain Generation 提出基于反馈空间搜索的模型空间推理方法，用于规划领域自动生成。 large language model
29 Squeeze Evolve: Unified Multi-Model Orchestration for Verifier-Free Evolution Squeeze Evolve：用于无验证器进化的统一多模型编排框架 multimodal
30 Towards Knowledgeable Deep Research: Framework and Benchmark 提出混合知识分析框架HKA，解决深度研究中结构化与非结构化知识融合问题。 multimodal

🔬 支柱二：RL算法与架构 (RL & Architecture) (15 篇)

#题目一句话要点标签🔗⭐
31 ASPECT:Analogical Semantic Policy Execution via Language Conditioned Transfer ASPECT：通过语言条件迁移实现模拟语义策略执行 reinforcement learning large language model language conditioned
32 WorldMAP: Bootstrapping Vision-Language Navigation Trajectory Prediction with Generative World Models WorldMAP：利用生成式世界模型引导视觉-语言导航轨迹预测 world model world models egocentric
33 Ads in AI Chatbots? An Analysis of How Large Language Models Navigate Conflicts of Interest 分析大型语言模型在利益冲突下的行为，揭示广告植入对用户福利的影响 reinforcement learning large language model
34 Multimodal Reasoning with LLM for Encrypted Traffic Interpretation: A Benchmark 提出BGTD基准和mmTraffic框架，用于加密流量的可解释多模态推理。 Mamba multimodal ✅
35 Investigation of Automated Design of Quantum Circuits for Imaginary Time Evolution Methods Using Deep Reinforcement Learning 利用深度强化学习自动设计量子电路，加速虚时演化算法 reinforcement learning deep reinforcement learning
36 Grounding Clinical AI Competency in Human Cognition Through the Clinical World Model and Skill-Mix Framework 提出临床世界模型和技能组合框架，弥合临床AI能力与人类认知之间的差距 world model world models
37 SUPERNOVA: Eliciting General Reasoning in LLMs with Reinforcement Learning on Natural Instructions SUPERNOVA：利用自然指令上的强化学习提升LLM的通用推理能力 reinforcement learning large language model ✅
38 Aligning Agents via Planning: A Benchmark for Trajectory-Level Reward Modeling 提出Plan-RewardBench，用于评估工具集成环境中轨迹级奖励模型的对齐能力。 reinforcement learning RLHF large language model
39 HiRO-Nav: Hybrid ReasOning Enables Efficient Embodied Navigation 提出HiRO-Nav以解决长时间导航任务中的推理效率问题 reinforcement learning multimodal
40 Beyond Stochastic Exploration: What Makes Training Data Valuable for Agentic Search 提出HiExp框架，提升Agentic Search中LLM推理效率与训练稳定性 reinforcement learning large language model
41 ReRec: Reasoning-Augmented LLM-based Recommendation Assistant via Reinforcement Fine-tuning 提出ReRec，通过强化微调增强LLM在推荐任务中的推理能力 reward shaping instruction following ✅
42 Mitigating Distribution Sharpening in Math RLVR via Distribution-Aligned Hint Synthesis and Backward Hint Annealing 提出DAHS和BHA，缓解数学RLVR中分布锐化问题，提升解题覆盖率。 reinforcement learning teacher-student
43 ASPECT:Analogical Semantic Policy Execution via Language Conditioned Transfer ASPECT：通过语言条件迁移实现模拟语义策略执行 reinforcement learning large language model language conditioned
44 RAMP: Hybrid DRL for Online Learning of Numeric Action Models 提出RAMP混合DRL算法，用于在线学习数值动作模型。 reinforcement learning deep reinforcement learning DRL
45 eBandit: Kernel-Driven Reinforcement Learning for Adaptive Video Streaming 提出eBandit以解决自适应视频流中的网络监测不足问题 reinforcement learning

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

#题目一句话要点标签🔗⭐
46 From Gaze to Guidance: Interpreting and Adapting to Users' Cognitive Needs with Multimodal Gaze-Aware AI Assistants 提出基于眼动追踪的多模态AI助手，提升用户认知能力。 egocentric multimodal

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
47 Securing Retrieval-Augmented Generation: A Taxonomy of Attacks, Defenses, and Future Directions 提出RAG安全威胁分类体系，分析攻击、防御与未来方向 manipulation large language model

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

#题目一句话要点标签🔗⭐
48 PSI: Shared State as the Missing Layer for Coherent AI-Generated Instruments in Personal AI Agents PSI：共享状态作为个人AI Agent中连贯AI生成工具的关键层 affordance

⬅️ 返回 cs.AI 首页 · 🏠 返回主页