| 一种新的分层强化学习方法 |
| |
| 引用本文： | 沈晶,顾国昌,刘海波.一种新的分层强化学习方法[J].计算机应用,2006,26(8):1938-1939. |
| |
| 作者姓名： | 沈晶 顾国昌 刘海波 |
| |
| 作者单位： | 哈尔滨工程大学,计算机科学与技术学院,黑龙江,哈尔滨,150001 |
| |
| 基金项目： | 哈尔滨工程大学校科研和教改项目 |
| |
| 摘 要： | 提出一种集成Option和MAXQ的分层强化学习新方法——OMQ，该方法以MAXQ为基本框架利用先验知识对任务进行人工分层和在线学习，集成Option方法对难以预先细分的子任务进行自动分层。以出租车问题为背景对OMQ学习算法进行了仿真与对比分析，实验结果表明，在任务环境不完全可知条件下，OMQ比Option和MAXQ更适用。
|
| 关 键 词： | 分层强化学习 Option MAXQ |
| 文章编号： | 1001-9081（2006）08-1938-02 |
| 收稿时间： | 2006-02-24 |
| 修稿时间： | 2006-02-24 |
New method of hierarchical reinforcement learning |
| |
| SHEN Jing,GU Guo-chang,LIU Hai-bo.New method of hierarchical reinforcement learning[J].journal of Computer Applications,2006,26(8):1938-1939. |
| |
| Authors: | SHEN Jing GU Guo-chang LIU Hai-bo |
| |
| Affiliation: | School of Computer Science and Technology, Harbin Engineering University, Harbin Heilongjiang 150001, China |
| |
| Abstract: | A novel method of hierarchical reinforcement learning which named OMQ was presented by integrating Options into MAXQ. In OMQ, MAXQ was used as the basic framework to design hierarchies experientially and learn online, and the Option was used to construct hierarchies automatically. The performance of OMQ was demonstrated in taxi domain and compared with Option and MAXQ. The simulation results show that the OMQ is more practical than Option and MAXQ in partially known environment. |
| |
| Keywords: | Option MAXQ |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
| 点击此处可从《计算机应用》浏览原始摘要信息 |
|
点击此处可从《计算机应用》下载全文 |
