通信学报 ›› 2015, Vol. 36 ›› Issue (2): 186-192.doi: 10.11959/j.issn.1000-436x.2015047

• 学术论文 • 上一篇    下一篇

支持合并的自适应tile coding算法

施梦宇1,刘全1,2,傅启明1   

  1. 1 苏州大学 计算机科学与技术学院,江苏 苏州 215006
    2 吉林大学 符号计算与知识工程教育部重点实验室,吉林 长春 130012
  • 出版日期:2015-02-25 发布日期:2017-06-27
  • 基金资助:
    国家自然科学基金资助项目;国家自然科学基金资助项目;江苏省自然科学基金资助项目

Mergeable adaptive tile coding method

Meng-yu SHI1,Quan LIU1,2,Qi-ming FU1   

  1. 1 School of Computer Science and Technology,Soochow University,Suzhou 215006,China
    2 Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education,Jilin University,Changchun 130012,China
  • Online:2015-02-25 Published:2017-06-27
  • Supported by:
    The National Natural Science Foundation of China;The National Natural Science Foundation of China;The National Science Foundation of Jiangsu Province

摘要:

针对自适应 tile coding 算法会产生多余划分的问题,提出一种支持合并的自适应 tile coding 算法——MATC。该算法能够消除传统自适应tile coding算法中产生的多余划分,进一步解决连续状态空间离散化的问题。将MATC算法应用于离散动作连续状态的Mountain Car问题上,实验结果表明,该算法在学习过程中能消除传统tile coding算法的误划分所产生的不良影响,更准确地自动调整划分的精度,并更快地收敛到最佳策略。

关键词: 连续空间, 离散化, 强化学习, 自适应, tilecoding

Abstract:

In order to solve many unnecessary division,mergence supported adaptive tile coding algorithm was presented which would eliminate the unnecessary division.Simulation is conducted on mountain car problem with discrete actions and continuous state space Results show that the proposed method can eliminate the influence of false division in the traditional tile coding method and achieve a more accurate adaptive partition of continuous state space.A higher convergence rate is achieved at the same time.

Key words: continuous space, discretization, reinforcement learning, adaptive; tile coding

No Suggested Reading articles found!