Improved behavior equivalence algorithm of multi-agent interactive dynamic influence diagrams

CORC > 厦门大学 > 信息技术－已发表论文

	Improved behavior equivalence algorithm of multi-agent interactive dynamic influence diagrams
	Tian, Le ; Luo, Jian ; Cao, Langcai ; Luo J(罗键) ; Cao LC(曹浪财)
刊名	http://dx.doi.org/10.13245/j.hust.140413
	2014
关键词	Artificial intelligence Forestry
英文摘要	The look-ahead search method was used to give a new method for determining approximate behavior equivalence. The method first determined whether the models were approximately behavior equivalent by comparing part of the solution (i.e. policy tree), then quickly clustered top-down and pruned the models that were approximately behavior equivalent. Next, the method used representative model to expand the interactive dynamic influence diagrams into flat dynamic influence diagrams. Finally, the flat dynamic influence diagrams were solved. The method reduces the storage space and the running time, thus improves the efficiency of the algorithm. The effectiveness of the proposed method was verified through experiments on multi-agent tiger and multi-agent concert problems.
语种	英语
出版者	Huazhong University of Science and Technology
内容类型	期刊论文
源URL	[http://dspace.xmu.edu.cn/handle/2288/92958]
专题	信息技术－已发表论文
推荐引用方式 GB/T 7714	Tian, Le,Luo, Jian,Cao, Langcai,et al. Improved behavior equivalence algorithm of multi-agent interactive dynamic influence diagrams[J]. http://dx.doi.org/10.13245/j.hust.140413,2014.
APA	Tian, Le,Luo, Jian,Cao, Langcai,罗键,&曹浪财.(2014).Improved behavior equivalence algorithm of multi-agent interactive dynamic influence diagrams.http://dx.doi.org/10.13245/j.hust.140413.
MLA	Tian, Le,et al."Improved behavior equivalence algorithm of multi-agent interactive dynamic influence diagrams".http://dx.doi.org/10.13245/j.hust.140413 (2014).