cs.AI（2025-12-22）

📊 共 22 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) (13 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (6) 支柱一：机器人控制 (Robot Control) (1) 支柱四：生成式动作 (Generative Motion) (1) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (13 篇)

#题目一句话要点标签🔗⭐
1 PENDULUM: A Benchmark for Assessing Sycophancy in Multimodal Large Language Models 提出PENDULUM基准，评估多模态大语言模型中的谄媚现象 large language model multimodal ✅
2 Understanding Chain-of-Thought in Large Language Models via Topological Data Analysis 利用拓扑数据分析理解大语言模型中的思维链 large language model chain-of-thought
3 FC-MIR: A Mobile Screen Awareness Framework for Intent-Aware Recommendation based on Frame-Compressed Multimodal Trajectory Reasoning 提出FC-MIR框架，通过帧压缩多模态轨迹推理实现意图感知的移动屏幕推荐。 large language model multimodal
4 The Epistemological Consequences of Large Language Models: Rethinking collective intelligence and institutional knowledge 大型语言模型对认知的影响：重新思考集体智能与机构知识 large language model
5 Clustering-based Transfer Learning for Dynamic Multimodal MultiObjective Evolutionary Algorithm 提出基于聚类迁移学习的动态多模态多目标进化算法，解决动态环境下的多模态优化问题。 multimodal
6 VIGOR+: Iterative Confounder Generation and Validation via LLM-CEVAE Feedback Loop VIGOR+：提出基于LLM-CEVAE反馈环路的迭代混淆因子生成与验证框架，解决因果推断中的隐藏混淆问题。 large language model
7 The Erasure Illusion: Stress-Testing the Generalization of LLM Forgetting Evaluation 提出Erasure Illusion框架，用于压力测试LLM遗忘评估的泛化能力。 large language model
8 Efficient Jailbreak Mitigation Using Semantic Linear Classification in a Multi-Staged Pipeline 提出一种基于语义线性分类的多阶段流水线，高效缓解大语言模型的越狱攻击。 large language model
9 A Dataset and Preliminary Study of Using GPT-5 for Code-change Impact Analysis 构建代码变更影响分析数据集，初步评估GPT-5在代码影响预测中的能力。 large language model
10 An Agentic Framework for Autonomous Materials Computation 提出基于Agent的材料计算框架，实现第一性原理计算的可靠自动化。 large language model
11 Causal-Guided Detoxify Backdoor Attack of Open-Weight LoRA Models 提出CBA：一种因果引导的LoRA模型解毒后门攻击方法 large language model
12 Observer, Not Player: Simulating Theory of Mind in LLMs through Game Observation 提出基于观察者模式的框架，通过石头剪刀布游戏评估LLM的心理理论能力 large language model
13 Population-Evolve: a Parallel Sampling and Evolutionary Method for LLM Math Reasoning 提出Population-Evolve，一种基于遗传算法的LLM数学推理优化方法 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🔗⭐
14 Training Multimodal Large Reasoning Models Needs Better Thoughts: A Three-Stage Framework for Long Chain-of-Thought Synthesis and Selection 提出SynSelect框架，为多模态大模型生成高质量长链推理训练数据。 reinforcement learning multimodal chain-of-thought
15 Tool-Augmented Hybrid Ensemble Reasoning with Distillation for Bilingual Mathematical Problem Solving 提出HERALD框架，融合工具增强、集成推理与知识蒸馏，提升双语数学问题求解能力 reinforcement learning distillation large language model
16 SafeMed-R1: Adversarial Reinforcement Learning for Generalizable and Robust Medical Reasoning in Vision-Language Models SafeMed-R1：用于视觉-语言模型中可泛化和鲁棒医学推理的对抗强化学习 reinforcement learning chain-of-thought
17 Helios: A Foundational Language Model for Smart Energy Knowledge Reasoning and Application Helios：面向智慧能源知识推理与应用的领域专用大语言模型 RLHF large language model
18 Can abstract concepts from LLM improve SLM performance? 利用LLM抽象概念提升SLM性能，实现推理时动态调整 distillation large language model
19 Learning General Policies with Policy Gradient Methods 提出基于图神经网络的策略梯度方法，学习可泛化的通用策略 reinforcement learning DRL

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
20 Generation of Programmatic Rules for Document Forgery Detection Using Large Language Models 利用大语言模型生成程序化规则，用于文档伪造检测 manipulation large language model

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
21 Towards Closed-Loop Embodied Empathy Evolution: Probing LLM-Centric Lifelong Empathic Motion Generation in Unseen Scenarios 提出基于LLM的终身情感动作生成框架，解决新场景下的情感动作泛化问题 motion generation

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

#题目一句话要点标签🔗⭐
22 Vibe Reasoning: Eliciting Frontier AI Mathematical Capabilities -- A Case Study on IMO 2025 Problem 6 提出Vibe Reasoning，提升AI在复杂数学问题上的推理能力 IMoS

⬅️ 返回 cs.AI 首页 · 🏠 返回主页