cs.AI（2026-04-27）

📊 共 25 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) (20 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (3) 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) (20 篇)

#题目一句话要点标签🔗⭐
1 Jailbreaking Frontier Foundation Models Through Intention Deception 提出基于意图欺骗的多轮对话攻击方法，破解前沿大语言模型 foundation model multimodal
2 Representational Curvature Modulates Behavioral Uncertainty in Large Language Models 研究表明表征曲率调节大型语言模型中的行为不确定性 large language model
3 Learning to Route Queries to Heads for Attention-based Re-ranking with Large Language Models 提出RouteHead，通过学习查询相关的head选择，提升基于LLM的注意力重排序效果。 large language model
4 Quantum Kernel Advantage over Classical Collapse in Medical Foundation Model Embeddings 利用量子核在医学基础模型嵌入上超越经典方法，提升胸部X光片保险分类性能 foundation model ✅
5 Layerwise Convergence Fingerprints for Runtime Misbehavior Detection in Large Language Models 提出层间收敛指纹LCF，用于大语言模型运行时异常行为检测。 large language model
6 Strategic Bidding in 6G Spectrum Auctions with Large Language Models 利用大语言模型在6G频谱拍卖中实现策略性竞标，提升资源利用率 large language model
7 An Information-Geometric Framework for Stability Analysis of Large Language Models under Entropic Stress 提出基于信息几何的框架，评估大语言模型在不确定性下的稳定性。 large language model
8 A2DEPT: Large Language Model-Driven Automated Algorithm Design via Evolutionary Program Trees A2DEPT：基于大语言模型与进化程序树的自动化算法设计 large language model
9 XGRAG: A Graph-Native Framework for Explaining KG-based Retrieval-Augmented Generation 提出XGRAG框架，通过图扰动解释知识图谱增强的检索增强生成 large language model
10 Defective Task Descriptions in LLM-Based Code Generation: Detection and Analysis 提出SpecValidator，用于检测LLM代码生成中任务描述的缺陷。 large language model
11 Leveraging LLMs for Multi-File DSL Code Generation: An Industrial Case Study 利用LLM生成多文件DSL代码：宝马工业案例研究 large language model
12 AgentWard: A Lifecycle Security Architecture for Autonomous AI Agents AgentWard：面向自主AI代理的全生命周期安全架构 large language model ✅
13 STELLAR-E: a Synthetic, Tailored, End-to-end LLM Application Rigorous Evaluator STELLAR-E：提出一种全自动的、可定制的LLM应用评测数据集生成框架。 large language model
14 Interoceptive machine framework: Toward interoception-inspired regulatory architectures in artificial intelligence 提出基于内感受的机器框架，旨在提升人工智能的自适应自主性。 embodied AI
15 Why AI Harms Can't Be Fixed One Identity at a Time: What 5300 Incident Reports Reveal About Intersectionality 揭示AI偏见：基于5300份事故报告的交叉性分析框架 large language model
16 GAMMAF: A Common Framework for Graph-Based Anomaly Monitoring Benchmarking in LLM Multi-Agent Systems GAMMAF：LLM多智能体系统中基于图的异常监控基准测试通用框架 large language model
17 MEMCoder: Multi-dimensional Evolving Memory for Private-Library-Oriented Code Generation MEMCoder：面向私有库代码生成的多维演化记忆框架 large language model
18 RefEvo: Agentic Design with Co-Evolutionary Verification for Agile Reference Model Generation RefEvo：利用协同进化验证的Agentic设计，加速敏捷参考模型生成 large language model
19 Defusing the Trigger: Plug-and-Play Defense for Backdoored LLMs via Tail-Risk Intrinsic Geometric Smoothing 提出TIGS：一种即插即用的后门大语言模型防御方法，无需额外数据和模型更新。 large language model
20 What Did They Mean? How LLMs Resolve Ambiguous Social Situations across Perspectives and Roles 研究表明大型语言模型倾向于消除社会情境中的歧义，而非保持不确定性。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (3 篇)

#题目一句话要点标签🔗⭐
21 MIMIC: A Generative Multimodal Foundation Model for Biomolecules MIMIC：用于生物分子的生成式多模态基础模型，实现跨模态生物分子状态建模与设计。 representation learning foundation model multimodal
22 Grounding Before Generalizing: How AI Differs from Humans in Causal Transfer 揭示LLM/VLM在因果迁移学习中对环境依赖性，与人类的抽象推理存在差距 reinforcement learning large language model multimodal
23 Aligning with Your Own Voice: Self-Corrected Preference Learning for Hallucination Mitigation in LVLMs 提出AVES-DPO框架以解决LVLM中的幻觉问题 preference learning DPO

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

#题目一句话要点标签🔗⭐
24 IntentVLM: Open-Vocabulary Intention Recognition through Forward-Inverse Modeling with Video-Language Models IntentVLM：利用视频-语言模型和前向-逆向建模实现开放词汇意图识别 open-vocabulary open vocabulary multimodal

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

#题目一句话要点标签🔗⭐
25 AgentPulse: A Continuous Multi-Signal Framework for Evaluating AI Agents in Deployment 提出AgentPulse框架以评估AI代理在实际部署中的表现 PULSE

⬅️ 返回 cs.AI 首页 · 🏠 返回主页