cs.AI（2025-02-19）

📊 共 20 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (13 🔗4) 支柱二：RL算法与架构 (RL & Architecture) (4) 支柱一：机器人控制 (Robot Control) (2) 支柱四：生成式动作 (Generative Motion) (1)

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

#	题目	一句话要点	标签	🔗	⭐
1	Benchmarking graph construction by large language models for coherence-driven inference	提出一种算法客观生成连贯性推理图，并评估LLM重建能力。	large language model
2	Lost in Sequence: Do Large Language Models Understand Sequential Recommendation?	提出LLM-SRec，通过知识蒸馏提升大语言模型在序列推荐中的性能	large language model	✅
3	Helix-mRNA: A Hybrid Foundation Model For Full Sequence mRNA Therapeutics	提出Helix-mRNA混合模型，用于优化全序列mRNA疗法，显著提升序列长度和参数效率。	foundation model	✅
4	Investigating Non-Transitivity in LLM-as-a-Judge	揭示LLM评判中的非传递性问题，提出基于循环赛和动态匹配的更可靠排序方法	large language model instruction following
5	LaVCa: LLM-assisted Visual Cortex Captioning	LaVCa：利用LLM辅助视觉皮层活动进行自然语言描述，提升脑活动理解	large language model
6	A Comprehensive Survey on Composed Image Retrieval	对组合图像检索（CIR）任务进行全面综述，为该领域研究提供及时概览。	multimodal	✅
7	A Mousetrap: Fooling Large Reasoning Models for Jailbreak with Chain of Iterative Chaos	Mousetrap：利用迭代混沌链破解大型推理模型的越狱攻击框架	large language model
8	Giving AI Personalities Leads to More Human-Like Reasoning	通过赋予AI人格提升其类人推理能力，解决完整推理谱问题	large language model
9	Improving LLM-powered Recommendations with Personalized Information	CoT-Rec：通过个性化信息增强LLM驱动的推荐系统	chain-of-thought	✅
10	Proving Olympiad Inequalities by Synergizing LLMs and Symbolic Reasoning	提出一种神经符号方法，结合LLM与符号推理，解决奥林匹克不等式证明难题。	large language model
11	A consensus set for the aggregation of partial rankings: the case of the Optimal Set of Bucket Orders Problem	提出OSBOP方法，通过生成排序集合解决排序聚合问题，提升结果多样性和适应性。	multimodal
12	Agentic AI Software Engineers: Programming with Trust	基于信任的Agentic AI软件工程师：利用LLM Agent提升软件工程自动化	large language model
13	Modeling Behavior Change for Multi-model At-Risk Students Early Prediction (extended version)	提出MCPD模型，融合多模态数据与变点检测，用于早期预测高危学生。	multimodal

🔬 支柱二：RL算法与架构 (RL & Architecture) (4 篇)

#	题目	一句话要点	标签	🔗	⭐
14	SPPD: Self-training with Process Preference Learning Using Dynamic Value Margin	提出SPPD框架，利用动态价值边际进行过程偏好学习，提升LLM的数理逻辑推理能力。	reinforcement learning preference learning DPO
15	Vision-Based Generic Potential Function for Policy Alignment in Multi-Agent Reinforcement Learning	提出基于视觉潜在函数的MARL策略对齐方法，提升智能体通用常识理解	reinforcement learning reward shaping large language model
16	Scoring Verifiers: Evaluating Synthetic Verification for Code and Reasoning	提出评估合成验证器的新基准，用于提升代码和推理能力。	reinforcement learning large language model
17	Secure Federated Data Distillation	提出安全联邦数据蒸馏框架SFDD，在保护隐私的同时实现数据集的去中心化蒸馏。	distillation

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#	题目	一句话要点	标签	🔗	⭐
18	Model Evolution Framework with Genetic Algorithm for Multi-Task Reinforcement Learning	提出基于遗传算法的模型演化框架MEGA，提升多任务强化学习性能	manipulation reinforcement learning
19	Exploiting Prefix-Tree in Structured Output Interfaces for Enhancing Jailbreak Attacking	提出AttackPrefixTree (APT)框架，针对结构化输出接口提升大语言模型越狱攻击成功率。	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	🔗	⭐
20	Think-Then-React: Towards Unconstrained Human Action-to-Reaction Generation	提出Think-Then-React框架，用于生成符合人类行为逻辑的反应动作	motion generation motion tokenizer egocentric

⬅️ 返回 cs.AI 首页 · 🏠 返回主页