cs.AI（2026-03-26）

📊 共 30 篇论文 | 🔗 2 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (19 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (8) 支柱一：机器人控制 (Robot Control) (3 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (19 篇)

#题目一句话要点标签🔗⭐
1 Can MLLMs Read Students' Minds? Unpacking Multimodal Error Analysis in Handwritten Math 提出 ScratchMath 基准，用于多模态大语言模型分析手写数学错误 large language model multimodal
2 RubricEval: A Rubric-Level Meta-Evaluation Benchmark for LLM Judges in Instruction Following RubricEval：针对LLM指令跟随能力评估的细粒度元评估基准 large language model instruction following
3 CSI-tuples-based 3D Channel Fingerprints Construction Assisted by MultiModal Learning 提出基于CSI元组和多模态学习的3D信道指纹构建框架，提升低空通信信道信息获取精度。 multimodal
4 Is Mathematical Problem-Solving Expertise in Large Language Models Associated with Assessment Performance? 研究表明LLM的数学问题解决能力与其评估学生解题步骤的准确性显著相关 large language model
5 A Gait Foundation Model Predicts Multi-System Health Phenotypes from 3D Skeletal Motion 提出步态基础模型，通过3D骨骼运动预测多系统健康表型 foundation model
6 Large Language Models as Optimization Controllers: Adaptive Continuation for SIMP Topology Optimization 提出基于大语言模型的拓扑优化自适应控制方法，提升SIMP算法性能。 large language model
7 AD-CARE: A Guideline-grounded, Modality-agnostic LLM Agent for Real-world Alzheimer's Disease Diagnosis with Multi-cohort Assessment, Fairness Analysis, and Reader Study AD-CARE：基于指南、模态无关的LLM智能体，用于真实世界阿尔茨海默病诊断 large language model multimodal
8 TopoPilot: Reliable Conversational Workflow Automation for Topological Data Analysis and Visualization TopoPilot：面向拓扑数据分析与可视化的可靠对话式工作流自动化框架 large language model
9 SliderQuant: Accurate Post-Training Quantization for LLMs SliderQuant：面向LLM的精确后训练量化框架，提升不同层量化精度。 large language model
10 WebTestBench: Evaluating Computer-Use Agents towards End-to-End Automated Web Testing WebTestBench：用于评估计算机使用代理的端到端自动化Web测试基准 large language model ✅
11 PIDP-Attack: Combining Prompt Injection with Database Poisoning Attacks on Retrieval-Augmented Generation Systems 提出PIDP-Attack，结合提示注入与数据库投毒攻击RAG系统 large language model
12 Trace2Skill: Distill Trajectory-Local Lessons into Transferable Agent Skills Trace2Skill：通过轨迹局部学习提炼可迁移的Agent技能 large language model
13 Factors Influencing the Quality of AI-Generated Code: A Synthesis of Empirical Evidence 系统性综述AI代码生成质量影响因素，揭示人机协作关键作用 large language model
14 ElephantBroker: A Knowledge-Grounded Cognitive Runtime for Trustworthy AI Agents ElephantBroker：用于可信AI代理的知识驱动认知运行时 large language model
15 Sparse Visual Thought Circuits in Vision-Language Models 探究视觉语言模型中稀疏视觉思维回路的可组合性与可控性 multimodal
16 From Stateless to Situated: Building a Psychological World for LLM-Based Emotional Support LEKIA 2.0：构建情境感知的LLM心理支持系统，解决多轮对话中的状态缺失问题 large language model
17 The Anatomy of Uncertainty in LLMs 提出LLM不确定性分解框架，提升模型可靠性并检测幻觉 large language model
18 Evaluating adaptive and generative AI-based feedback and recommendations in a knowledge-graph-integrated programming learning system 提出融合知识图谱的自适应与生成式AI编程学习反馈框架，提升学习效果。 large language model
19 LogitScope: A Framework for Analyzing LLM Uncertainty Through Information Metrics LogitScope：通过信息度量分析LLM不确定性的轻量级框架 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (8 篇)

#题目一句话要点标签🔗⭐
20 A Wireless World Model for AI-Native 6G Networks 提出无线世界模型WWM，用于AI原生6G网络中预测无线信道时空演化。 world model world models Joint-Embedding Predictive Architecture
21 Macroscopic Characteristics of Mixed Traffic Flow with Deep Reinforcement Learning Based Automated and Human-Driven Vehicles 基于深度强化学习的自动驾驶车辆提升混合交通流的宏观性能 reinforcement learning deep reinforcement learning DRL
22 R-C2: Cycle-Consistent Reinforcement Learning Improves Multimodal Reasoning 提出R-C2框架以解决多模态推理中的一致性问题 reinforcement learning multimodal
23 Modernising Reinforcement Learning-Based Navigation for Embodied Semantic Scene Graph Generation 改进基于强化学习的导航，用于具身语义场景图生成 reinforcement learning world model world models
24 EcoThink: A Green Adaptive Inference Framework for Sustainable and Accessible Agents EcoThink：提出绿色自适应推理框架，降低LLM能耗并提升可访问性 distillation large language model chain-of-thought
25 Reinforcement learning for quantum processes with memory 提出基于乐观最大似然估计的强化学习算法，解决量子过程中的记忆问题。 reinforcement learning
26 A Public Theory of Distillation Resistance via Constraint-Coupled Reasoning Architectures 提出基于约束耦合推理架构的蒸馏抵抗理论框架，提升模型安全性。 distillation
27 Training the Knowledge Base through Evidence Distillation and Write-Back Enrichment 提出WriteBack-RAG，通过知识蒸馏和回写增强训练RAG系统的知识库。 distillation

🔬 支柱一：机器人控制 (Robot Control) (3 篇)

#题目一句话要点标签🔗⭐
28 Beyond Content Safety: Real-Time Monitoring for Reasoning Vulnerabilities in Large Language Models 提出推理安全监控器，实时检测大语言模型推理过程中的漏洞 manipulation large language model chain-of-thought
29 From Manipulation to Mistrust: Explaining Diverse Micro-Video Misinformation for Robust Debunking in the Wild 提出WildFakeBench和FakeAgent，用于检测和解释微视频中的多样化虚假信息。 manipulation multimodal ✅
30 Evaluating Language Models for Harmful Manipulation 提出评估框架，研究AI在公共政策、金融和健康领域中的有害操纵行为。 manipulation

⬅️ 返回 cs.AI 首页 · 🏠 返回主页