期刊文献+
任意字段
题名或关键词
题名
关键词
文摘
作者
第一作者
机构
刊名
分类号
参考文献
作者简介
基金资助
栏目信息

强化学习中基于定性模型的知识传递方法 被引量：1

Knowledge Transfer Method Based on the Qualitative Model in Reinforcement Learning
下载PDF
导出
摘要 本文提出一种基于定性模糊网络的强化学习知识传递方法。该方法通过建立系统的定性模型,并用定性模糊网络抽取基于定性动作的次优策略的共同特征获得与系统参数无关知识。这些知识能有效描述参数值不同的系统所具有的共同控制规律,加快在新参数值的系统中强化学习的收敛速度。 This paper proposes a new reinforcement learning knowledge transfer method based on a qualitative model.The method defines the qualitative model and extracts the common features of the sub-optimal policy to obtain knowledge by qualitative fuzzy networks.The knowledge can represent the common features of the tasks with different parameters.The convergence can be accelerated by the knowledge unrelated to the parameters.
作者 黄晗文 郑宇
机构地区 湖南大学计算机与通信学院 湖南工业职业技术学院信息工程系 北京交通大学计算机与信息技术学院
出处 《计算机工程与科学》 CSCD 北大核心 2011年第6期118-124,共7页 Computer Engineering & Science
基金 省教育厅科学研究项目(09C1134)
关键词 强化学习 定性模型 知识传递 reinforcement learning qualitative model knowledge transfer
分类号 TP18 [自动化与计算机技术—控制理论与控制工程]
  • 相关文献

参考文献15

  • 1Sherstove A, Stone P. Improving Action Selection in MDPs via Knowledge Transfer[C]//Proc of the Twentieth National Conf on Artificial Intelligence, 2005 : 1024- 1029.
  • 2Liu Y, Stone P. Value-Function Based Transfer for Reinforcement Learning Using Structure Mapping[C]//Proc of the National Conf on Artificial Intelligence, 2006:415-420.
  • 3Taylor M, Jong N, Stone P. Transferring Instances for Model- Based Reinforcement Learning[C]//Proc of the Adaptive Learning Agents and Multi-Agent Systems, 2008:488-.505.
  • 4Mehta N, Natarajan S, Tadepalli P, et al. Transfer in Vari able Reward Hierarchical Reinforcement Learning[J]. Machine Learning, 2008, 73(3):289 -312.
  • 5Konidaris G, Barto A. Building Portable Options: Skill Transfer in Reinforcement Learning [C]//Proc of the Int' l Joint Conf on Artificial Intelligence, 2007:895 -900.
  • 6Torrey L, J. Shavlik W, Walker T, et al. Rule Extraction for Transfer Learning[M] // Rule Extraction from Support Vector Machines, 2008:67 -82.
  • 7Perkins T J, Precup D. Using Options for Knowledge Transfer in Reinforcement Learning[R]. Technical Report UM CS-1999-034, University of Massachusetts, 1999.
  • 8Lazarie A, Restelli M, Bonarini A. Tranfer of Samples in Batch Reinforcement Learning[C]//Proc of the Int'l Conf on Machine Learning,2008 :544 -551.
  • 9Rosenstein M T, Marx Z, Kaelbling L, et al. To Transfer or Not To Transfer[C]//Proe of NIPS Workshop on Inductive Transfer, 2005.
  • 10Sutton R, Barto A. Reinforcement Learning: An Introduction[M]. Cambridge, MA:MIT Press, 1998.

同被引文献95

  • 1Laura RAY.Hierarchical state-abstracted and socially augmented Q-Learning for reducing complexity in agent-based learning[J].控制理论与应用（英文版）,2011,9(3):440-450. 被引量：2
  • 2MURRAY R M,ASTROM K M,BODY S P,et al.Future directions in control in an information-rich world[J].IEEE Control Systems Magazine,2003,23 (2):20-23.
  • 3WIERING M,OTTERLO M V.Reinforcement learning state-of-the-art[M].Berlin:Springer-Verlag,2012:3-42.
  • 4SUTTON R S.Learning to predict by the methods of temporal differences[J].Machine Learning,1988,3(1):9-44.
  • 5CHEN Xingguo,GAO Yang,WANG Ruili.Online selective kernel-based temporal difference learning[J].IEEE Transactions on Neural Networks and Learning Systems,2013,24(12):1944-1956.
  • 6ZOU Bin,ZHANG Hai,XU Zongben.Learning from uniformly ergodic Markov chains[J].Journal of Complexity,2009,25(2):188-200.
  • 7YU Huizhen,BERTSEKAS D P.Convergence results for some temporal difference methods based on least squares[J].IEEE Transactions on Automatic Control,2009,54(7):1515-1531.
  • 8WATKINS C,DAYAN P.Q-learning[J].Machine Learning,1992,8(3):279-292.
  • 9CHEN Chunlin,DONG Daoyi,LI Hanxiong.Fidelitybased probabilistic Q-learning for control of quantum systems[J].IEEE Transactions on Neural Networks and Learning Systems,2014,25(5):920-933.
  • 10RUMMERY G,NIRANJAN M.On-line Q-learning using connectionist systems[D].Cambridge:University of Cambridge,1994.

引证文献1

二级引证文献14

计算机工程与科学
2011年 第6期

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部