Parallel recombinative reinforcement learning: A genetic approach

Likas, A; Blekas, K; Stafylopatis, A

dc.contributor.author	Likas, A	en
dc.contributor.author	Blekas, K	en
dc.contributor.author	Stafylopatis, A	en
dc.date.accessioned	2014-03-01T01:44:49Z
dc.date.available	2014-03-01T01:44:49Z
dc.date.issued	1996	en
dc.identifier.issn	03341860	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/24494
dc.relation.uri	http://www.scopus.com/inward/record.url?eid=2-s2.0-0029698326&partnerID=40&md5=3d1ff6f0bdacbf3a8dde56486a5e41c0	en
dc.subject	Genetic algorithms	en
dc.subject	Graph partitioning	en
dc.subject	Hybrid technique	en
dc.subject	Optimization	en
dc.subject	Parallel implementation	en
dc.subject	Reinforcement learning	en
dc.subject.other	Artificial intelligence	en
dc.subject.other	Genetic algorithms	en
dc.subject.other	Graph theory	en
dc.subject.other	Learning systems	en
dc.subject.other	Probability	en
dc.subject.other	Random processes	en
dc.subject.other	Hybrid technique	en
dc.subject.other	Parallel implementation	en
dc.subject.other	Optimization	en
dc.title	Parallel recombinative reinforcement learning: A genetic approach	en
heal.type	journalArticle	en
heal.publicationDate	1996	en
heal.abstract	A technique is presented that is suitable for function optimization in high-dimensional binary domains. The method allows an efficient parallel implementation and is based on the combination of genetic algorithms and reinforcement learning schemes. More specifically, a population of probability vectors is considered, each member corresponding to a reinforcement learning optimizer. Each probability vector represents the adaptable parameters of a team of stochastic units whose binary outputs provide a point of the function state space. At each step of the proposed technique the population members are updated according to a reinforcement learning rule and then recombined in a manner analogous to traditional genetic algorithm operation. Special care is devoted to ensuring the desirable properties of sustained exploration capability and sustained population diversity. The method has been tested on the graph partitioning problem in comparison with other techniques under two different types of fitness evaluation yielding very promising results.	en
heal.journalName	Journal of Intelligent Systems	en
dc.identifier.volume	6	en
dc.identifier.issue	2	en
dc.identifier.spage	145	en
dc.identifier.epage	169	en