cs.AI（2025-06-14）

📊 共 24 篇论文 | 🔗 3 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (19 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (4 🔗1) 支柱一：机器人控制 (Robot Control) (1)

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

#	题目	一句话要点	标签	🔗	⭐
1	CMI-Bench: A Comprehensive Benchmark for Evaluating Music Instruction Following	CMI-Bench：一个全面的音乐指令跟随评估基准，用于评估音频-文本大语言模型在音乐信息检索任务中的性能。	large language model instruction following
2	Information Suppression in Large Language Models: Auditing, Quantifying, and Characterizing Censorship in DeepSeek	提出审计框架，揭示DeepSeek大语言模型中的信息抑制现象	large language model chain-of-thought
3	DeepSeq: High-Throughput Single-Cell RNA Sequencing Data Labeling via Web Search-Augmented Agentic Generative AI Foundation Models	DeepSeq：利用Web搜索增强的Agentic生成式AI基础模型进行高通量单细胞RNA测序数据标记	foundation model
4	A Comprehensive Survey of Deep Research: Systems, Methodologies, and Applications	对基于AI的深度研究系统：系统、方法与应用的全面综述	large language model foundation model multimodal	✅
5	MEraser: An Effective Fingerprint Erasure Approach for Large Language Models	提出MEraser以有效去除大语言模型的指纹	large language model
6	Automated Heuristic Design for Unit Commitment Using Large Language Models	提出基于大语言模型的FunSearch方法，用于自动设计电力系统机组组合方案	large language model
7	DinoCompanion: An Attachment-Theory Informed Multimodal Robot for Emotionally Responsive Child-AI Interaction	DinoCompanion：基于依恋理论的多模态机器人，用于情感响应式儿童-AI互动	multimodal
8	MALM: A Multi-Information Adapter for Large Language Models to Mitigate Hallucination	提出MALM多信息适配器，利用多图学习缓解大语言模型幻觉问题	large language model
9	CORONA: A Coarse-to-Fine Framework for Graph-based Recommendation with Large Language Models	提出CORONA框架，利用大语言模型进行图推荐的粗到细候选过滤。	large language model
10	Step-by-Step Reasoning Attack: Revealing 'Erased' Knowledge in Large Language Models	提出 Sleek 攻击，揭示大语言模型中基于逐步推理的知识擦除漏洞	large language model
11	Model Merging for Knowledge Editing	提出基于模型融合的知识编辑框架，提升LLM在序列编辑中的性能并保持通用能力	large language model foundation model	✅
12	Behavioral Generative Agents for Energy Operations	提出基于生成式Agent的能源运营消费者行为建模方法	large language model
13	Evaluating AI Alignment in Eleven LLMs through Output-Based Analysis and Human Benchmarking	PAPERS框架：通过输出分析和人类基准评估11个LLM中的AI对齐程度	large language model
14	Graph of Verification: Structured Verification of LLM Reasoning with Directed Acyclic Graphs	提出GoV框架，通过有向无环图结构化验证LLM推理过程，提升验证的适应性和精度。	large language model
15	SheetMind: An End-to-End LLM-Powered Multi-Agent Framework for Spreadsheet Automation	SheetMind：基于LLM的多智能体电子表格自动化框架	large language model
16	The Foundation Cracks: A Comprehensive Study on Bugs and Testing Practices in LLM Libraries	首个LLM库缺陷与测试实践的综合研究，揭示API误用是主要问题	large language model
17	The Budget AI Researcher and the Power of RAG Chains	提出基于RAG链的Budget AI Researcher框架，用于生成更具体、更有趣的科研idea。	large language model
18	QGuard:Question-based Zero-shot Guard for Multi-modal LLM Safety	提出QGuard以解决多模态LLM安全问题	large language model
19	The SWE-Bench Illusion: When State-of-the-Art LLMs Remember Instead of Reason	揭示SWE-Bench的局限性：大型语言模型可能记忆而非推理	large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (4 篇)

#	题目	一句话要点	标签	🔗	⭐
20	MM-R5: MultiModal Reasoning-Enhanced ReRanker via Reinforcement Learning for Document Retrieval	提出MM-R5，通过强化学习增强多模态文档检索的推理重排序能力	reinforcement learning multimodal instruction following	✅
21	Ghost Policies: A New Paradigm for Understanding and Learning from Failure in Deep Reinforcement Learning	提出Ghost Policies，通过增强现实可视化DRL失败轨迹，促进人机协同学习。	reinforcement learning deep reinforcement learning DRL
22	Trust-MARL: Trust-Based Multi-Agent Reinforcement Learning Framework for Cooperative On-Ramp Merging Control in Heterogeneous Traffic Flow	提出Trust-MARL框架，解决异构交通流中匝道汇入控制问题	reinforcement learning penetration
23	Theoretical Tensions in RLHF: Reconciling Empirical Success with Inconsistencies in Social Choice Theory	理论社会选择矛盾：调和RLHF经验成功与理论不一致性	reinforcement learning RLHF

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#	题目	一句话要点	标签	🔗	⭐
24	Feeling Machines: Ethics, Culture, and the Rise of Emotional AI	情感AI伦理、文化与发展：跨学科视角下的机遇与挑战	manipulation

⬅️ 返回 cs.AI 首页 · 🏠 返回主页