cs.AI（2025-12-12）

📊 共 25 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) 支柱八：物理动画 (Physics-based Animation) (1) 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 The Forecast Critic: Leveraging Large Language Models for Poor Forecast Identification 提出Forecast Critic，利用大语言模型识别不合理的预测结果，提升零售业务效率。 large language model
2 Log Anomaly Detection with Large Language Models via Knowledge-Enriched Fusion EnrichLog：利用知识增强融合的大语言模型进行日志异常检测 large language model
3 AI-MASLD Metabolic Dysfunction and Information Steatosis of Large Language Models in Unstructured Clinical Narratives 揭示大语言模型在非结构化临床叙述中存在的“AI-MASLD”代谢功能障碍 large language model
4 LOOPRAG: Enhancing Loop Transformation Optimization with Retrieval-Augmented Large Language Models LOOPRAG：利用检索增强的大语言模型提升循环变换优化效果 large language model
5 Semantic search for 100M+ galaxy images using AI-generated captions AION-Search：利用AI生成描述，实现对1.4亿张星系图像的语义搜索。 foundation model multimodal ✅
6 Rethinking Jailbreak Detection of Large Vision Language Models with Representational Contrastive Scoring 提出RCS框架，利用表征对比评分提升大型视觉语言模型的越狱攻击检测能力 multimodal ✅
7 Instruction-Tuning Open-Weight Language Models for BPMN Model Generation 提出InstruBPM，通过指令微调开源大语言模型生成高质量BPMN流程模型。 large language model
8 Super Suffixes: Bypassing Text Generation Alignment and Guard Models Simultaneously 提出Super Suffixes攻击绕过文本生成对齐与防御模型，并提出DeltaGuard防御方法。 large language model
9 From Verification Burden to Trusted Collaboration: Design Goals for LLM-Assisted Literature Reviews 针对LLM辅助文献综述，提出可信协作的设计目标与框架，提升研究效率。 large language model
10 AI Benchmark Democratization and Carpentry 倡导AI基准测试民主化与基准测试构建，应对AI评估的快速演进与实际部署风险。 large language model
11 How AI Agents Follow the Herd of AI? Network Effects, History, and Machine Optimism 研究AI智能体在网络效应博弈中的决策行为，揭示历史数据对群体决策的影响 large language model
12 EmeraldMind: A Knowledge Graph-Augmented Framework for Greenwashing Detection 提出EmeraldMind框架，利用知识图谱增强的检索增强生成技术，用于检测企业漂绿行为。 large language model
13 Persistent Backdoor Attacks under Continual Fine-Tuning of LLMs P-Trojan：一种在LLM持续微调下保持持久性的后门攻击方法 large language model
14 Towards Privacy-Preserving Code Generation: Differentially Private Code Language Models 提出差分隐私CodeLLM，解决代码生成模型中的隐私泄露问题。 large language model
15 General-purpose AI models can generate actionable knowledge on agroecological crop protection 利用通用AI模型为农业生态作物保护生成可执行知识 large language model
16 AGAPI-Agents: An Open-Access Agentic AI Platform for Accelerated Materials Design on AtomGPT.org AGAPI-Agents：AtomGPT.org上用于加速材料设计的开放式Agentic AI平台 large language model ✅
17 FutureWeaver: Planning Test-Time Compute for Multi-Agent Systems with Modularized Collaboration FutureWeaver：通过模块化协作规划多智能体系统的测试时计算资源 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 Evolutionary Reinforcement Learning based AI tutor for Socratic Interdisciplinary Instruction 提出ERL4SIIP，基于演化强化学习的AI导师用于苏格拉底式跨学科教学 reinforcement learning PPO large language model
19 Towards Trustworthy Multi-Turn LLM Agents via Behavioral Guidance 提出基于行为引导的多轮LLM Agent框架，提升任务可靠性和可验证性 reinforcement learning large language model
20 A-LAMP: Agentic LLM-Based Framework for Automated MDP Modeling and Policy Generation A-LAMP：基于Agentic LLM的自动化MDP建模与策略生成框架 reinforcement learning large language model
21 Reliable Policy Iteration: Performance Robustness Across Architecture and Environment Perturbations 提出可靠策略迭代(RPI)，提升深度强化学习在环境和架构扰动下的鲁棒性。 PPO TD3
22 Three methods, one problem: Classical and AI approaches to no-three-in-line 首次系统比较整数线性规划与AI方法求解No-Three-In-Line问题 reinforcement learning PPO

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
23 amc: The Automated Mission Classifier for Telescope Bibliographies 提出amc，利用大型语言模型自动分类望远镜文献，提升天文研究效率。 PULSE large language model

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

#题目一句话要点标签🔗⭐
24 A Scalable Multi-GPU Framework for Encrypted Large-Model Inference Cerium：用于加密大模型推理的可扩展多GPU框架 OMOMO

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

#题目一句话要点标签🔗⭐
25 PHANTOM: Progressive High-fidelity Adversarial Network for Threat Object Modeling PHANTOM：一种用于威胁对象建模的渐进式高保真对抗网络 feature matching

⬅️ 返回 cs.AI 首页 · 🏠 返回主页