Exploration in Gradient-Based Reinforcement Learning

dc.creator	Meuleau, Nicolas
dc.creator	Peshkin, Leonid
dc.creator	Kim, Kee-Eung
dc.date	2004-10-04T14:37:39Z
dc.date	2004-10-04T14:37:39Z
dc.date	2001-04-03
dc.date.accessioned	2013-10-09T02:42:44Z
dc.date.available	2013-10-09T02:42:44Z
dc.date.issued	2013-10-09
dc.identifier	AIM-2001-003
dc.identifier	http://hdl.handle.net/1721.1/6076
dc.identifier.uri	http://koha.mediu.edu.my:8181/xmlui/handle/1721
dc.description	Gradient-based policy search is an alternative to value-function-based methods for reinforcement learning in non-Markovian domains. One apparent drawback of policy search is its requirement that all actions be 'on-policy'; that is, that there be no explicit exploration. In this paper, we provide a method for using importance sampling to allow any well-behaved directed exploration policy during learning. We show both theoretically and experimentally that using this method can achieve dramatic performance improvements.
dc.format	5594043 bytes
dc.format	516972 bytes
dc.format	application/postscript
dc.format	application/pdf
dc.language	en_US
dc.relation	AIM-2001-003
dc.title	Exploration in Gradient-Based Reinforcement Learning

Files in this item

Files	Size	Format	View
There are no files associated with this item.