cs.AI（2026-04-21）

📊 共 20 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) (11 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (7) 支柱一：机器人控制 (Robot Control) (2)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (11 篇)

#题目一句话要点标签🔗⭐
1 SafetyALFRED: Evaluating Safety-Conscious Planning of Multimodal Large Language Models SafetyALFRED：评估多模态大语言模型在具身环境中安全意识规划能力 large language model multimodal ✅
2 A-MAR: Agent-based Multimodal Art Retrieval for Fine-Grained Artwork Understanding 提出A-MAR，基于Agent的多模态艺术品检索框架，用于细粒度的艺术品理解。 large language model multimodal ✅
3 ProjLens: Unveiling the Role of Projectors in Multimodal Model Safety ProjLens揭示投影层在多模态模型安全性中的作用，助力后门攻击分析与防御。 large language model multimodal
4 GRASPrune: Global Gating for Budgeted Structured Pruning of Large Language Models GRASPrune：面向大语言模型预算约束的全局门控结构化剪枝 large language model
5 Multimodal Transformer for Sample-Aware Prediction of Metal-Organic Framework Properties EXIT：结合XRD的多模态Transformer用于金属有机框架的样本感知属性预测 multimodal
6 SimDiff: Depth Pruning via Similarity and Difference SimDiff：通过相似性和差异性进行深度剪枝，提升LLM部署效率 large language model
7 Do Agents Dream of Root Shells? Partial-Credit Evaluation of LLM Agents in Capture The Flag Challenges DeepRed：一个用于评估LLM智能体在CTF挑战中表现的基准测试框架，并提出部分信用评分方法。 large language model
8 Streamliners for Answer Set Programming 利用大语言模型为解答集编程生成Streamliner约束，提升求解效率。 large language model
9 DP-FlogTinyLLM: Differentially private federated log anomaly detection using Tiny LLMs 提出DP-FLogTinyLLM，用于在保护隐私的联邦环境中进行日志异常检测。 large language model
10 Towards Scalable Lifelong Knowledge Editing with Selective Knowledge Suppression LightEdit：通过选择性知识抑制实现可扩展的终身知识编辑 large language model
11 DW-Bench: Benchmarking LLMs on Data Warehouse Graph Topology Reasoning DW-Bench：用于评估LLM在数据仓库图拓扑推理能力的新基准 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (7 篇)

#题目一句话要点标签🔗⭐
12 DT2IT-MRM: Debiased Preference Construction and Iterative Training for Multimodal Reward Modeling DT2IT-MRM：通过去偏好构建与迭代训练提升多模态奖励模型性能 RLHF large language model multimodal
13 OLLM: Options-based Large Language Models OLLM：基于选项的大语言模型，提升数学推理任务的可控性和效率。 reinforcement learning policy learning large language model
14 Cyber Defense Benchmark: Agentic Threat Hunting Evaluation for LLMs in SecOps 提出网络安全防御基准，评估LLM在SecOps中威胁狩猎任务的表现 reinforcement learning large language model TAMP
15 Multi-modal Reasoning with LLMs for Visual Semantic Arithmetic 提出SAri-RFT，增强LVLM在视觉语义算术任务中的推理能力，应用于机器人领域。 reinforcement learning large language model
16 Reasoning-Aware AIGC Detection via Alignment and Reinforcement 提出REVEAL框架，通过对齐和强化推理能力提升AIGC文本检测性能 reinforcement learning large language model
17 Reinforcement Learning Improves LLM Accuracy and Reasoning in Disease Classification from Radiology Reports 利用强化学习提升LLM在放射报告疾病分类中的准确性和推理能力 reinforcement learning
18 Reasoning Structure Matters for Safety Alignment of Reasoning Models AltTrain：通过改变推理结构实现推理模型安全对齐 reinforcement learning reward design

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
19 Detecting Data Contamination in Large Language Models 评估黑盒成员推理攻击在大型语言模型数据污染检测中的可靠性 manipulation large language model
20 Large Language Models Exhibit Normative Conformity 揭示大语言模型中的规范性顺从，为LLM多智能体系统决策提供安全保障。 manipulation large language model

⬅️ 返回 cs.AI 首页 · 🏠 返回主页