logo
科言猫学术研究的AI总结
language
手机端手机端
  • Phone Feature微信小程序
简体中文简体中文
notice消息通知
博客博客
delete
arrow
返回

Diffusion Dynamic Model for Unsupervised Reinforcement Learning

delete2026-01-01
delete0
PRE
AI
R
Ran Chen
胡小亮 封面图
胡小亮 (Xiaoliang Hu)
崔
崔振 (Zhen Cui)
L
Luying Wu
T
Tong Zhang *
DOI：10.1007/978-981-95-5693-9_1delete
delete原文链接
delete原文求助
delete分享
delete收藏
摘要

摘要

En 中文
无监督强化学习（URL）使智能体能够高效适应新任务，但依赖于鲁棒的环境建模。传统世界模型增强了探索能力，但易出现单步错误，导致分布偏移。为解决此挑战，我们引入扩散动态模型（DDM），将扩散模型整合到无监督强化学习的环境建模组件中。在预训练阶段，DDM基于当前观测和动作生成未来观测。这些生成观测用于设计内在奖励函数，激励智能体探索多样性和高不确定性状态。在微调阶段，DDM利用其生成能力进行数据增强。通过生成多样、高质量合成数据，DDM扩展训练数据集，使智能体更好地泛化并更快适应特定任务环境。我们的方法在URLB基准测试的三个领域和十二个下游任务中得到验证，展现出比现有方法更优越的探索能力和适应性。
Keyword:
Diffusion dynamic model
Unsupervised reinforcement learning
Data augmentation

期刊

P
PATTERN RECOGNITION AND COMPUTER VISION, PRCV 2025, PT II
IF:
0
论文数:
27
被引数:
0

机构

N
nanjing university of science & technology
学者数:
1.6K
论文数: 544
被引数: 0
B
beijing normal university
学者数:
5.6K
论文数: 2.2K
被引数: 0
引用论文

引用论文

Diffusion for World Modeling: Visual Details Matter in Atari
errAdvances in Neural Information Processing Systems 37
IF0
err2024-01-01
err0
PREAI
errAlonso,Eloi; Fleuret,François; Jelley,Adam; Kanervisto,Anssi; Micheli,Vincent; Pearce,Tim; Storkey,Amos
err分享
err收藏
Curiosity-Driven Exploration by Self-Supervised Prediction通过自我监督的预测进行好奇心驱动的探索
err2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
IF0
err2017-07-01
err0
errOAAI
errDeepak Pathak; Pulkit Agrawal; Alexei A. Efros; Trevor Darrell
err分享
err收藏
Multi-Scenario Combination Based on Multi-Agent Reinforcement Learning to Optimize the Advertising Recommendation System基于多智能体强化学习的多场景组合优化广告推荐系统
err2024 5th International Conference on Artificial Intelligence and Electromechanical Automation (AIEA)
IF0
err2024-06-14
err0
errOAAI
errYang Zhao; Chang Zhou; Jin Cao; Yi Zhao; Shaobo Liu; Chiyu Cheng; Xingchen Li
err分享
err收藏
Scaling Robot Learning with Semantically Imagined Experience
errRobotics: Science and Systems XIX
IF0
err2023-07-10
err0
errOAAI
errTianhe Yu; Ted Xiao; Jonathan Tompson; Austin Stone; Su Wang; Anthony Brohan; Jaspiar Singh; Clayton Tan; Dee M; Jodilyn Peralta; Karol Hausman; Brian Ichter; Fei Xia
err分享
err收藏
Reinforcement Learning: An Introduction
errIEEE Transactions on Neural Networks
IF0
err1998-09-01
err0
PREAI
errR.S. Sutton; A.G. Barto
err分享
err收藏
Deep Reinforcement Learning for Robotics: A Survey of Real-World Successes
errProceedings of the AAAI Conference on Artificial Intelligence
IF0
err
err0
PREAI
errTang,Chen; Abbatematteo,Ben; Hu,Jiaheng; Chandra,Rohan; Martín-Martín,Roberto; Stone,Peter
err分享
err收藏
没有更多内容