[1]王云,韩伟.一种基于划分和集成思想的多智能体强化学习[J].南京师范大学学报(工程技术版),2008,08(04):059-62.
　Wang Yun,Han Wei.An Multiagent Reinforcement Learning Based on Partition and Integration[J].Journal of Nanjing Normal University(Engineering and Technology),2008,08(04):059-62.
点击复制

一种基于划分和集成思想的多智能体强化学习
分享到：

南京师范大学学报（工程技术版）[ISSN:1006-6977/CN:61-1281/TN]

卷:
08卷
期数:
2008年04期
页码:
059-62
栏目:
出版日期:
2008-12-30

文章信息/Info

Title:
An Multiagent Reinforcement Learning Based on Partition and Integration
作者:
王云;韩伟;
南京财经大学信息工程学院, 江苏南京210046
Author(s):
Wang Yun; Han Wei
Information Science and Engineering College,Nanjing University of Financial and Economics,Nanjing 210046,China
关键词:
多智能体系统;  强化学习;  状态空间划分
Keywords:
mu ltiagent system;  re inforcem ent learn ing;  state- space partition
分类号:
TP301
摘要:
针对Q学习状态空间非常大,导致收敛速度非常慢的问题,利用智能体在不同样本上分类性能不同,提出了基于样本的学习误差对样本空间进行划分,充分发掘了样本和智能体的匹配关系.以带障碍物的格子世界作为仿真环境,表明该算法提高了在线学习性能.
Abstract:
To counte r for the prob lem of slow ly convergence of Q leaning w hen com e ing to large state-space, the paper pu ts forw ard an a lgo rithm w hich divide the sta tes space acco rd ing to learn ing e rrors. The basic idea o f our algor ithm is to d iscover the m atch ing re lationship be tw een ag ents and the sub- space o f sta tes space. The sim ulations in g rids w ith b locks ind icate that the algorithm perform s betterw hen com e ing to on- line learning.

参考文献/References:

[ 1] 刘海涛, 洪炳熔, 朴松昊, 等. 不确定环境下基于进化算法的强化学习[ J]. 电子学报, 2006, 7( 34): 1 356-1 360.
Liu H aitao, H ong B ingrong, Pu Songhao, et a.l Evolutionary a lgor ithm based re inforcem ent learn ing in the uncerta in env ironm ents[ J]. A cta E lectron ica S in ica. 2006( 34) 7: 1 356-1 360. ( in Ch inese)
[ 2] 韩伟, 陈优广, 姜昌华. 基于内省推理的多agent在线学习新方法[ J] . 模式识别与人工智能, 2007, 20( 2) : 254-260.
H anW e,i Chen Youguang, Jiang Changhua. An Interna l- In fe rence BasedM ultiagent Learning M ethod[ J]. Pattern Recognition & Artificia l Inte lligence. 2007, 20( 2): 254-260. ( in Ch inese)
[ 3] 韩伟. 基于情节序列训练的电子市场智能定价算法[ J] . 计算机工程与应用, 2007, 43( 6): 17-19.
H anW e.i Intelligent pr ic ing a lgo rithm based on mu ltiagent lea rning [ J]. Com pute r Eng ineer ing and Applica tions. 2007( 43)6: 17-19. ( in Chinese)
[ 4] 文益民, 杨旸, 吕宝粮. 集成学习算法在增量学习中的应用研究[ J] . 计算机研究与发展, 2005, 42: 222-227.
W en Y im in, Y ang Yang, L?Bao liang. Rescearch o f the app lication ensem ble lea rning algor ithm s to increm enta l lea rn ing [ J].Jou rnal of Com puter Research and Developm ent, 2005, 42: 222-227. ( in Ch inese)

备注/Memo

备注/Memo:
基金项目: 国家自然科学基金( 70802025)资助项目.
通讯联系人: 王云, 讲师, 研究方向: 电子商务、人工智能. E-m ail:dallashw@ gm ail.com
更新日期/Last Update: 2013-04-24