cs.AI（2025-12-19）

📊 共 21 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (12 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (5) 支柱一：机器人控制 (Robot Control) (3) 支柱四：生成式动作 (Generative Motion) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (12 篇)

#题目一句话要点标签🔗⭐
1 Towards Explainable Conversational AI for Early Diagnosis with Large Language Models 提出基于LLM的对话式AI，用于早期诊断并提升可解释性 large language model chain-of-thought
2 Quantifying Laziness, Decoding Suboptimality, and Context Degradation in Large Language Models 量化大语言模型的惰性、次优解码和上下文退化现象 large language model instruction following
3 Attention Distance: A Novel Metric for Directed Fuzzing with Large Language Models 提出注意力距离以解决现有模糊测试中的逻辑关系缺失问题 large language model ✅
4 SWE-Bench++: A Framework for the Scalable Generation of Software Engineering Benchmarks from Open-Source Repositories SWE-Bench++：一个可扩展的软件工程基准测试框架，从开源仓库自动生成测试用例。 large language model
5 Holistic Evaluation of State-of-the-Art LLMs for Code Generation 全面评估大型语言模型在代码生成任务中的性能表现 large language model
6 Rethinking Multi-Agent Intelligence Through the Lens of Small-World Networks 利用小世界网络优化多智能体系统通信拓扑，提升共识稳定性 large language model
7 Specification and Detection of LLM Code Smells 定义并检测LLM代码异味，提升软件系统质量 large language model
8 LLM-based Behaviour Driven Development for Hardware Design 提出基于LLM的硬件设计行为驱动开发方法，提升测试验证效率 large language model
9 Eidoku: A Neuro-Symbolic Verification Gate for LLM Reasoning via Structural Constraint Satisfaction Eidoku：通过结构约束满足实现LLM推理的神经符号验证门 large language model
10 UmniBench: Unified Understand and Generation Model Oriented Omni-dimensional Benchmark 提出 UmniBench，用于全面评估统一多模态模型的理解、生成和编辑能力。 multimodal
11 PILAR: Personalizing Augmented Reality Interactions with LLM-based Human-Centric and Trustworthy Explanations for Daily Use Cases PILAR：利用LLM提供个性化AR交互解释，提升日常使用场景的用户体验和信任度 large language model
12 QMBench: A Research Level Benchmark for Quantum Materials Research QMBench：用于评估大语言模型在量子材料研究中能力的基准测试 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🔗⭐
13 Large Language Models as Pokémon Battle Agents: Strategic Play and Content Generation 利用大型语言模型作为宝可梦对战智能体，实现战略决策与内容生成 reinforcement learning large language model
14 MMRAG-RFT: Two-stage Reinforcement Fine-tuning for Explainable Multi-modal Retrieval-augmented Generation 提出MMRAG-RFT，通过两阶段强化学习微调实现可解释的多模态检索增强生成。 reinforcement learning large language model multimodal
15 Unifying Causal Reinforcement Learning: Survey, Taxonomy, Algorithms and Applications 综述因果强化学习：统一视角，分类算法与应用 reinforcement learning representation learning
16 AlignDP: Hybrid Differential Privacy with Rarity-Aware Protection for LLMs AlignDP：针对LLM的混合差分隐私方法，通过稀有感知保护防止知识泄露。 distillation large language model
17 About Time: Model-free Reinforcement Learning with Timed Reward Machines 提出基于时序奖励机的免模型强化学习方法，解决时序约束下的非马尔可夫奖励问题。 reinforcement learning

🔬 支柱一：机器人控制 (Robot Control) (3 篇)

#题目一句话要点标签🔗⭐
18 HydroGym: A Reinforcement Learning Platform for Fluid Dynamics HydroGym：用于流体动力学的强化学习平台，加速流动控制研究。 manipulation reinforcement learning
19 Accelerating Multi-modal LLM Gaming Performance via Input Prediction and Mishit Correction 提出基于输入预测和误差校正的多模态LLM游戏加速框架，提升实时控制性能。 humanoid MPC world model
20 Securing Agentic AI Systems -- A Multilayer Security Framework 提出MAAIS框架，解决Agentic AI系统在全生命周期中的安全问题。 manipulation

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
21 q3-MuPa: Quick, Quiet, Quantitative Multi-Parametric MRI using Physics-Informed Diffusion Models 提出q3-MuPa：基于物理信息扩散模型的快速、静音、定量多参数磁共振成像方法。 physics-informed diffusion

⬅️ 返回 cs.AI 首页 · 🏠 返回主页