cs.AI（2025-09-14）

📊 共 22 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) (16 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (5 🔗1) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (16 篇)

#题目一句话要点标签🔗⭐
1 Teaching LLMs to Plan: Logical Chain-of-Thought Instruction Tuning for Symbolic Planning 提出PDDL-Instruct框架，通过逻辑链式思维指令微调提升LLM的符号规划能力 large language model chain-of-thought
2 LibEMER: A novel benchmark and algorithms library for EEG-based Multimodal Emotion Recognition LibEMER：用于脑电多模态情感识别的基准测试与算法库 multimodal
3 Beyond Autoregression: An Empirical Study of Diffusion Large Language Models for Code Generation 首个扩散大语言模型代码生成实证研究，探索其在长代码理解和生成方面的潜力。 large language model ✅
4 Evalet: Evaluating Large Language Models by Fragmenting Outputs into Functions Evalet：通过将LLM输出分解为功能片段来评估其性能，提升评估可解释性。 large language model
5 Evolution of Kernels: Automated RISC-V Kernel Optimization with Large Language Models EoK：利用大语言模型自动优化RISC-V内核，解决参考资料稀缺问题 large language model
6 Tractable Asymmetric Verification for Large Language Models via Deterministic Replicability 提出基于确定性可复制性的LLM非对称验证框架，降低验证成本。 large language model
7 AIssistant: An Agentic Approach for Human--AI Collaborative Scientific Work on Reviews and Perspectives in Machine Learning AIssistant：一种用于人机协作完成机器学习综述和展望的 Agentic 方法 multimodal
8 Multi-Modal Sensing Aided mmWave Beamforming for V2V Communications with Transformers 提出基于Transformer的多模态融合毫米波波束赋形方法，降低V2V通信开销。 multimodal
9 LLMAP: LLM-Assisted Multi-Objective Route Planning with User Preferences LLMAP：基于LLM辅助的多目标个性化路线规划系统 large language model
10 Securing AI Agents: Implementing Role-Based Access Control for Industrial Applications 提出基于角色的访问控制框架，保障工业AI Agent安全 large language model
11 MAPGD: Multi-Agent Prompt Gradient Descent for Collaborative Prompt Optimization 提出MAPGD以解决单一轨迹优化的局限性问题 large language model
12 Decoding Plastic Toxicity: An Intelligent Framework for Conflict-Aware Relational Metapath Extraction from Scientific Abstracts 提出一种基于LLM的冲突感知关系元路径提取框架，用于解析塑料毒性。 large language model
13 VideoAgent: Personalized Synthesis of Scientific Videos 提出VideoAgent，通过多智能体框架个性化生成科学视频 multimodal
14 Difficulty-Aware Agentic Orchestration for Query-Specific Multi-Agent Workflows 提出难度感知Agent编排，为特定查询动态生成多Agent工作流 large language model
15 Beyond IVR Touch-Tones: Customer Intent Routing using LLMs 提出基于LLM的IVR意图路由方法，解决传统语音交互系统用户体验差的问题 large language model
16 Free-MAD: Consensus-Free Multi-Agent Debate 提出Free-MAD，解决多智能体辩论中一致性需求和误差传播问题 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🔗⭐
17 Omni-CLST: Error-aware Curriculum Learning with guided Selective chain-of-Thought for audio question answering 提出Omni-CLST框架，利用课程学习和选择性思维链提升音频问答性能。 curriculum learning multimodal chain-of-thought
18 Trading-R1: Financial Trading with LLM Reasoning via Reinforcement Learning 提出Trading-R1，利用强化学习提升LLM在金融交易中的推理能力。 reinforcement learning instruction following ✅
19 Mamba Outpaces Reformer in Stock Prediction with Sentiments from Top Ten LLMs 利用大型语言模型情感分析，Mamba在股票预测中超越Reformer Mamba large language model
20 Prompts to Proxies: Emulating Human Preferences via a Compact LLM Ensemble 提出偏好重构理论以解决人类偏好模拟问题 representation learning large language model
21 An Entropy-Guided Curriculum Learning Strategy for Data-Efficient Acoustic Scene Classification under Domain Shift 提出熵引导的课程学习策略，解决声场景分类中的数据稀疏和域偏移问题 curriculum learning

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
22 Your Compiler is Backdooring Your Model: Understanding and Exploiting Compilation Inconsistency Vulnerabilities in Deep Learning Compilers 揭示深度学习编译器漏洞：未经修改的编译器可引入模型后门 manipulation

⬅️ 返回 cs.AI 首页 · 🏠 返回主页