有限规划水平自适应Markov决策过程的参数决策

应用科学学报 ›› 2000, Vol. 18 ›› Issue (4): 335-339.

有限规划水平自适应Markov决策过程的参数决策

李江洪, 韩正之

上海交通大学智能工程研究所, 上海 200030

收稿日期:1999-11-17 修回日期:2000-01-17 出版日期:2000-12-31 发布日期:2000-12-31
作者简介:李江洪(1970-),男,湖南长沙人,博士生.
基金资助:
国家自然科学基金资助项目(69874025)

Parameter Decision Making in Adaptive Markov Decision Process with Finite Planning Horizon

LI Jiang-hong, HAN Zheng-zhi

Institute of Intelligence Engineering, Shanghai Jiaotong University, Shanghai 200030, China

Received:1999-11-17 Revised:2000-01-17 Online:2000-12-31 Published:2000-12-31

摘要/Abstract

摘要： 针对现有Markov决策过程自适应决策方法仅研究无限规划水平自适应决策的不足,提出了一种有限规划水平Markov决策过程自适应决策算法.算法的基本思想是运用Bayes理论对未知系统进行"学习",并且在每次决策时以最大概率保证实际决策为最优决策.最后用仿真结果表明了算法的有效性.

关键词: Markov决策过程, 自适应决策, Bayes原理

Abstract: An algorithm is proposed for adaptive MDP with finite planning horizon by reason of the fact that all current algorithms only consider adaptive MDP with infinite planning horizon. Bayes principle is applied to learn an unknown system; and for every decision the probability that the actual decision equals the optimal decision is maximized. Simulation results demonstrate the validity of the new algorithm.

Key words: Markov decision process (MDP), adaptive decision making, Bayes principle

中图分类号:

TP217.2

李江洪, 韩正之. 有限规划水平自适应Markov决策过程的参数决策[J]. 应用科学学报, 2000, 18(4): 335-339.

LI Jiang-hong, HAN Zheng-zhi. Parameter Decision Making in Adaptive Markov Decision Process with Finite Planning Horizon[J]. Journal of Applied Sciences, 2000, 18(4): 335-339.