cs.LG（2025-03-21）

📊 共 25 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) (16 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (7) 支柱一：机器人控制 (Robot Control) (1) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (16 篇)

#	题目	一句话要点	标签	🔗	⭐
1	Large Language Models Can Verbatim Reproduce Long Malicious Sequences	大型语言模型易受后门攻击，可精确复现长恶意序列	large language model
2	Large Language Model Compression via the Nested Activation-Aware Decomposition	提出嵌套激活感知分解(NSVD)方法，用于高效压缩大型语言模型。	large language model
3	Enhanced Smart Contract Reputability Analysis using Multimodal Data Fusion on Ethereum	提出基于多模态数据融合的智能合约信誉分析方法，提升以太坊生态信任度。	multimodal
4	Lie Detector: Unified Backdoor Detection via Cross-Examination Framework	提出统一的后门检测框架以解决安全风险问题	large language model multimodal
5	Fairness-Driven LLM-based Causal Discovery with Active Learning and Dynamic Scoring	提出基于LLM、主动学习和动态评分的因果发现框架，提升公平性分析效率。	large language model
6	Improving Quantization with Post-Training Model Expansion	提出后训练模型扩展方法，在量化LLM时提升模型质量并降低体积。	large language model
7	Variance Control via Weight Rescaling in LLM Pre-training	提出LIR初始化与TVR方差控制，提升LLM预训练性能并降低量化难度	large language model	✅
8	LEMMA: Learning from Errors for MatheMatical Advancement in LLMs	LEMMA：通过从错误中学习提升LLM的数学推理能力	large language model
9	FactSelfCheck: Fact-Level Black-Box Hallucination Detection for LLMs	FactSelfCheck：一种用于LLM的事实级黑盒幻觉检测方法	large language model
10	TreeSynth: Synthesizing Diverse Data from Scratch via Tree-Guided Subspace Partitioning	TreeSynth：通过树引导的子空间划分从零合成多样化数据	large language model	✅
11	Deterministic AI Agent Personality Expression through Standard Psychological Diagnostics	通过标准心理学诊断，实现AI Agent确定性人格表达	large language model
12	TRACE: Time SeRies PArameter EffiCient FinE-tuning	TRACE：面向时间序列基础模型的高效参数微调方法	foundation model
13	V-Seek: Accelerating LLM Reasoning on Open-hardware Server-class RISC-V Platforms	V-Seek：加速LLM在开源RISC-V服务器平台上的推理	large language model
14	Improving the End-to-End Efficiency of Offline Inference for Multi-LLM Applications Based on Sampling and Simulation	提出SamuLLM框架，通过采样和模拟提升多LLM应用离线推理效率。	large language model
15	Understanding Bias Reinforcement in LLM Agents Debate	提出DReaMAD框架，通过多样化推理和优化提示缓解LLM Agent辩论中的偏差强化问题	large language model
16	Enabling Global, Human-Centered Explanations for LLMs:From Tokens to Interpretable Code and Test Generation	提出CodeQ框架，实现代码大模型全局、以人为本的可解释性分析	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	🔗	⭐
17	Optimizing 2D+1 Packing in Constrained Environments Using Deep Reinforcement Learning	提出基于深度强化学习的2D+1约束环境下的排样优化方法	reinforcement learning deep reinforcement learning DRL
18	A Statistical Theory of Contrastive Learning via Approximate Sufficient Statistics	通过近似充分统计量建立对比学习的统计理论框架，提升表征学习效果	contrastive learning foundation model
19	Sparse Logit Sampling: Accelerating Knowledge Distillation in LLMs	提出基于重要性采样的稀疏Logit蒸馏方法，加速LLM知识蒸馏。	distillation large language model
20	Causally Aligned Curriculum Learning	提出因果对齐课程学习方法，解决强化学习中未观测混淆因素下的课程设计问题。	reinforcement learning curriculum learning
21	Generative Modeling of Class Probability for Multi-Modal Representation Learning	提出CALM模型，利用类概率分布生成建模实现多模态表征学习	representation learning contrastive learning
22	Efficient Knowledge Distillation via Curriculum Extraction	提出高效知识蒸馏方法以解决大规模训练挑战	distillation
23	Curriculum RL meets Monte Carlo Planning: Optimization of a Real World Container Management Problem	结合课程学习RL与蒙特卡洛规划，优化现实世界集装箱管理问题	reinforcement learning PPO curriculum learning

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#	题目	一句话要点	标签	🔗	⭐
24	End-to-End Deep Learning for Real-Time Neuroimaging-Based Assessment of Bimanual Motor Skills	提出一种端到端深度学习框架，用于实时神经影像评估双手动技能。	bi-manual

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

#	题目	一句话要点	标签	🔗	⭐
25	A Thorough Assessment of the Non-IID Data Impact in Federated Learning	联邦学习中非独立同分布数据影响的全面评估与量化分析	spatiotemporal

⬅️ 返回 cs.LG 首页 · 🏠 返回主页