cs.AI（2025-09-11）

📊 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) (17 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (5) 支柱五：交互与反应 (Interaction & Reaction) (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) (17 篇)

#	题目	一句话要点	标签	🔗	⭐
1	A Modular and Multimodal Generative AI Framework for Urban Building Energy Data: Generating Synthetic Homes	提出模块化多模态生成AI框架，用于生成城市建筑能源数据，合成住宅信息。	multimodal
2	Boosting Embodied AI Agents through Perception-Generation Disaggregation and Asynchronous Pipeline Execution	Auras：通过解耦感知-生成和异步流水线执行提升具身智能体性能	embodied AI
3	LoCoBench: A Benchmark for Long-Context Large Language Models in Complex Software Engineering	提出LoCoBench，用于评估长上下文LLM在复杂软件工程中的能力。	large language model	✅
4	Quality Assessment of Tabular Data using Large Language Models and Code Generation	提出基于大语言模型和代码生成的表格数据质量评估框架	large language model
5	On Integrating Large Language Models and Scenario-Based Programming for Improving Software Reliability	结合大语言模型与场景编程提升软件可靠性	large language model
6	DP-FedLoRA: Privacy-Enhanced Federated Fine-Tuning for On-Device Large Language Models	提出DP-FedLoRA，增强设备端LLM联邦微调的隐私保护。	large language model
7	Vibe Check: Understanding the Effects of LLM-Based Conversational Agents' Personality and Alignment on User Perceptions in Goal-Oriented Tasks	研究LLM对话Agent人格表达与用户匹配度对目标导向任务用户感知的影响	large language model
8	LLMs as Agentic Cooperative Players in Multiplayer UNO	利用LLM作为UNO多人游戏中具有能动性的合作玩家	large language model
9	Towards a Common Framework for Autoformalization	提出自动形式化通用框架，促进不同领域AI系统交叉融合	large language model
10	The Illusion of Diminishing Returns: Measuring Long Horizon Execution in LLMs	揭示LLM长程执行能力：单步精度提升带来任务完成长度的指数级增长	large language model
11	TORSO: Template-Oriented Reasoning Towards General Tasks	提出TORSO：面向模板推理，无需人工样本即可提升LLM在通用任务上的表现	large language model
12	Towards Adaptive ML Benchmarks: Web-Agent-Driven Construction, Domain Expansion, and Metric Optimization	提出TAM Bench，一个基于Web Agent驱动的自适应机器学习基准，用于评估LLM在端到端ML任务中的能力。	large language model
13	LightAgent: Production-level Open-source Agentic AI Framework	提出LightAgent：一个生产级开源Agentic AI框架，旨在简化多智能体系统部署。	large language model	✅
14	Jupiter: Enhancing LLM Data Analysis Capabilities via Notebook and Inference-Time Value-Guided Search	Jupiter：通过Notebook和推理时值引导搜索增强LLM数据分析能力	large language model	✅
15	Character-Level Perturbations Disrupt LLM Watermarks	提出基于字符级扰动的LLM水印移除攻击，揭示现有水印方案的脆弱性	large language model
16	Towards Confidential and Efficient LLM Inference with Dual Privacy Protection	CMIF：面向LLM推理的双重隐私保护框架，兼顾效率与安全性	large language model
17	Strategic Tradeoffs Between Humans and AI in Multi-Agent Bargaining	对比人类、LLM和贝叶斯智能体在多智能体议价中的策略权衡	large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (5 篇)

#	题目	一句话要点	标签	🔗	⭐
18	Curriculum-Based Multi-Tier Semantic Exploration via Deep Reinforcement Learning	提出基于课程学习的多层语义探索深度强化学习方法，提升具身智能体在未知环境中的探索效率。	reinforcement learning deep reinforcement learning DRL
19	How well can LLMs provide planning feedback in grounded environments?	评估LLM在具身环境中提供规划反馈的能力，揭示其优势与局限	policy learning reward design large language model
20	Tree-OPO: Off-policy Monte Carlo Tree-Guided Advantage Optimization for Multistep Reasoning	Tree-OPO：利用蒙特卡洛树搜索引导的优势优化多步推理	reinforcement learning policy learning large language model
21	SWE-Effi: Re-Evaluating Software AI Agent System Effectiveness Under Resource Constraints	SWE-Effi：在资源约束下重新评估软件AI Agent系统的有效性	reinforcement learning large language model
22	Adaptive Knowledge Distillation using a Device-Aware Teacher for Low-Complexity Acoustic Scene Classification	提出基于设备感知教师的自适应知识蒸馏方法，用于低复杂度声场景分类	distillation

🔬 支柱五：交互与反应 (Interaction & Reaction) (1 篇)

#	题目	一句话要点	标签	🔗	⭐
23	ENSI: Efficient Non-Interactive Secure Inference for Large Language Models	ENSI：面向大语言模型的高效非交互安全推理框架	OMOMO large language model

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

#	题目	一句话要点	标签	🔗	⭐
24	Mind Meets Space: Rethinking Agentic Spatial Intelligence from a Neuroscience-inspired Perspective	提出神经科学启发的Agentic空间智能框架，提升智能体在3D环境中的推理能力	egocentric multimodal

⬅️ 返回 cs.AI 首页 · 🏠 返回主页