cs.AI（2026-03-24）

📊 共 22 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) (15 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (6 🔗1) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (15 篇)

#题目一句话要点标签🔗⭐
1 Optimizing Small Language Models for NL2SQL via Chain-of-Thought Fine-Tuning 通过思维链微调优化小型语言模型，提升NL2SQL任务性能 large language model chain-of-thought
2 KARMA: Knowledge-Action Regularized Multimodal Alignment for Personalized Search at Taobao 提出KARMA框架，解决LLM在淘宝个性化搜索中知识与行为的对齐问题 large language model multimodal
3 Can Large Language Models Reason and Optimize Under Constraints? 评估大语言模型在约束条件下推理和优化能力，应用于电力系统优化 large language model
4 Reliable Classroom AI via Neuro-Symbolic Multimodal Reasoning 提出NSCR神经符号框架，用于构建可靠的多模态课堂AI系统 multimodal
5 AgriPestDatabase-v1.0: A Structured Insect Dataset for Training Agricultural Large Language Model 构建农业害虫知识库并微调轻量级LLM，为农业领域提供边缘设备决策支持 large language model
6 Robust Safety Monitoring of Language Models via Activation Watermarking 提出激活水印方法，提升大语言模型在对抗攻击下的安全监控鲁棒性 large language model
7 ReqFusion: A Multi-Provider Framework for Automated PEGS Analysis Across Software Domains ReqFusion：一个多LLM供应商框架，用于跨软件领域自动化PEGS需求分析 large language model
8 Beyond Preset Identities: How Agents Form Stances and Boundaries in Generative Societies 提出CMASE框架，研究生成式社会中Agent的立场形成与边界构建 large language model ✅
9 Leveraging LLMs and Social Media to Understand User Perception of Smartphone-Based Earthquake Early Warnings 利用大型语言模型和社交媒体分析用户对智能手机地震预警的感知 large language model
10 PERMA: Benchmarking Personalized Memory Agents via Event-Driven Preference and Realistic Task Environments PERMA：通过事件驱动偏好和真实任务环境评估个性化记忆代理 large language model ✅
11 Can an LLM Detect Instances of Microservice Infrastructure Patterns? MicroPAD利用LLM检测微服务架构模式实例，性能受模式特征影响 large language model
12 DBAutoDoc: Automated Discovery and Documentation of Undocumented Database Schemas via Statistical Analysis and Iterative LLM Refinement DBAutoDoc：通过统计分析和迭代LLM优化自动发现和文档化未文档化的数据库模式 large language model
13 JFTA-Bench: Evaluate LLM's Ability of Tracking and Analyzing Malfunctions Using Fault Trees JFTA-Bench：提出故障树文本表示，评估大语言模型在故障追踪与分析中的能力。 large language model
14 ProGRank: Probe-Gradient Reranking to Defend Dense-Retriever RAG from Corpus Poisoning 提出ProGRank，通过探针梯度重排序防御RAG中的语料库投毒攻击 large language model
15 Beyond Binary Correctness: Scaling Evaluation of Long-Horizon Agents on Subjective Enterprise Tasks 提出LH-Bench，用于评估长程Agent在主观企业任务中的表现，超越二元正确性。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (6 篇)

#题目一句话要点标签🔗⭐
16 CoMaTrack: Competitive Multi-Agent Game-Theoretic Tracking with Vision-Language-Action Models 提出CoMaTrack：基于竞争博弈的多智能体视觉-语言-动作跟踪框架 reinforcement learning imitation learning vision-language-action ✅
17 Improving Safety Alignment via Balanced Direct Preference Optimization 提出B-DPO，通过平衡偏好优化解决LLM安全对齐中的过拟合问题 reinforcement learning RLHF DPO
18 Describe-Then-Act: Proactive Agent Steering via Distilled Language-Action World Models 提出Dillo，通过蒸馏语言-动作世界模型实现主动Agent控制。 world model distillation large language model
19 MemCollab: Cross-Agent Memory Collaboration via Contrastive Trajectory Distillation MemCollab：通过对比轨迹蒸馏实现跨Agent的记忆协同 distillation large language model
20 Dynamical Systems Theory Behind a Hierarchical Reasoning Model 提出基于连续动力系统的Contraction Mapping Model，解决复杂推理任务中递归网络训练不稳定的问题。 latent dynamics large language model
21 Evaluating LLM-Based Test Generation Under Software Evolution 评估软件演化下基于LLM的测试生成：揭示其对语义变化的敏感性 SAC large language model

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
22 Chain-of-Authorization: Internalizing Authorization into Large Language Models via Reasoning Trajectories 提出Chain-of-Authorization框架，通过推理轨迹将授权机制内化于大语言模型中 manipulation large language model

⬅️ 返回 cs.AI 首页 · 🏠 返回主页