All learning is local: Multi-agent learning in global reward games

Please use this identifier to cite or link to this item: http://dspace.mediu.edu.my:8181/xmlui/handle/1721.1/3851

Full metadata record

DC Field	Value	Language
dc.creator	Chang, Yu-Han	-
dc.creator	Ho, Tracey	-
dc.creator	Kaelbling, Leslie P.	-
dc.date	2003-12-13T18:55:17Z	-
dc.date	2003-12-13T18:55:17Z	-
dc.date	2004-01	-
dc.date.accessioned	2013-10-09T02:32:49Z	-
dc.date.available	2013-10-09T02:32:49Z	-
dc.date.issued	2013-10-09	-
dc.identifier	http://hdl.handle.net/1721.1/3851	-
dc.identifier.uri	http://koha.mediu.edu.my:8181/xmlui/handle/1721	-
dc.description	In large multiagent games, partial observability, coordination, and credit assignment persistently plague attempts to design good learning algorithms. We provide a simple and efficient algorithm that in part uses a linear system to model the world from a single agent’s limited perspective, and takes advantage of Kalman filtering to allow an agent to construct a good training signal and effectively learn a near-optimal policy in a wide variety of settings. A sequence of increasingly complex empirical tests verifies the efficacy of this technique.	-
dc.description	Singapore-MIT Alliance (SMA)	-
dc.format	1408858 bytes	-
dc.format	application/pdf	-
dc.language	en_US	-
dc.relation	Computer Science (CS);	-
dc.subject	Kalman filtering	-
dc.subject	multi-agent systems	-
dc.subject	Q-learning	-
dc.subject	reinforcement learning	-
dc.title	All learning is local: Multi-agent learning in global reward games	-
dc.type	Article	-
Appears in Collections:	MIT Items

Files in This Item:

There are no files associated with this item.

DSpace JSPUI