cs.LG（2025-04-18）

📊 共 17 篇论文

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (10) 支柱一：机器人控制 (Robot Control) (3) 支柱二：RL算法与架构 (RL & Architecture) (3) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (10 篇)

#	题目	一句话要点	标签	🔗	⭐
1	Towards End-to-End Network Intent Management with Large Language Models	利用大语言模型实现5G/6G网络端到端意图管理	large language model
2	Designing a reliable lateral movement detector using a graph foundation model	利用图基础模型设计可靠的横向移动检测器	foundation model
3	Are you SURE? Enhancing Multimodal Pretraining with Missing Modalities through Uncertainty Estimation	SURE：通过不确定性估计增强缺失模态多模态预训练模型	multimodal
4	CAOTE: KV Cache Selection for LLMs via Attention Output Error-Based Token Eviction	CAOTE：基于Attention输出误差的KV缓存选择，提升LLM在资源受限设备上的性能。	large language model
5	Large Language Bayes	提出Large Language Bayes，利用大语言模型从非正式描述中构建贝叶斯模型并进行推理。	large language model
6	DP2Unlearning: An Efficient and Guaranteed Unlearning Framework for LLMs	DP2Unlearning：一种高效且有保障的LLM可遗忘学习框架	large language model
7	Scaling sparse feature circuit finding for in-context learning	利用稀疏自编码器，扩展稀疏特征电路发现方法，解析上下文学习机制。	large language model
8	Gradual Binary Search and Dimension Expansion : A general method for activation quantization in LLMs	提出基于梯度二分搜索和维度扩展的量化方法，实现LLM激活的低比特量化。	large language model
9	One Jump Is All You Need: Short-Cutting Transformers for Early Exit Prediction with One Jump to Fit All Exit Levels	提出One-Jump-Fits-All低秩捷径，显著降低Transformer模型早期退出的参数成本。	large language model
10	STAMP Your Content: Proving Dataset Membership via Watermarked Rephrasings	STAMP：通过水印式复述验证数据集在LLM预训练语料中的成员关系	large language model

🔬 支柱一：机器人控制 (Robot Control) (3 篇)

#	题目	一句话要点	标签	🔗	⭐
11	AutoAdv: Automated Adversarial Prompting for Multi-Turn Jailbreaking of Large Language Models	AutoAdv：自动化对抗提示框架，用于多轮破解大型语言模型	manipulation large language model
12	A Model-Based Approach to Imitation Learning through Multi-Step Predictions	提出基于多步预测的模型模仿学习框架，提升泛化性和鲁棒性	model predictive control imitation learning behavior cloning
13	Bounded and Uniform Energy-based Out-of-distribution Detection for Graphs	NODESAFE：通过有界能量和缓解logit偏移提升图神经网络的OOD节点检测	manipulation

🔬 支柱二：RL算法与架构 (RL & Architecture) (3 篇)

#	题目	一句话要点	标签	🔗	⭐
14	Not All Rollouts are Useful: Down-Sampling Rollouts in LLM Reinforcement Learning	提出PODS以解决LLM强化学习中的计算与内存不对称问题	reinforcement learning large language model
15	Personalizing Exposure Therapy via Reinforcement Learning	提出基于强化学习的暴露疗法个性化方法，用于虚拟现实环境下的蜘蛛恐惧症治疗。	reinforcement learning
16	CacheFormer: High Attention-Based Segment Caching	CacheFormer：提出基于分段缓存的高效Transformer长文本处理方法	SSM state space model

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

#	题目	一句话要点	标签	🔗	⭐
17	Stratify: Rethinking Federated Learning for Non-IID Data through Balanced Sampling	Stratify：通过平衡采样重思考非独立同分布数据下的联邦学习	OMOMO

⬅️ 返回 cs.LG 首页 · 🏠 返回主页