Show simple item record

dc.contributor.authorJacob, D.
dc.contributor.authorPolani, D.
dc.contributor.authorNehaniv, C.L.
dc.date.accessioned2008-06-10T10:02:48Z
dc.date.available2008-06-10T10:02:48Z
dc.date.issued2004
dc.identifier.citationJacob , D , Polani , D & Nehaniv , C L 2004 , ' Improving Learning for Embodied Agents in Dynamic--Environments by State Factorisation ' , Procs TAROS , vol. 2004 .
dc.identifier.issn1744-8050
dc.identifier.otherPURE: 96535
dc.identifier.otherPURE UUID: e27d8fd9-a9fe-4421-bde0-831159bc4e49
dc.identifier.otherdspace: 2299/2074
dc.identifier.urihttp://hdl.handle.net/2299/2074
dc.description.abstractA new reinforcement learning algorithm designed--specifically for robots and embodied systems--is described. Conventional reinforcement learning methods intended for learning general tasks suffer from a number of disadvantages in this domain including slow learning speed, an inability--to generalise between states, reduced performance--in dynamic environments, and a lack of scalability. Factor-Q, the new algorithm, uses factorised state and action, coupled with multiple structured rewards, to address these issues. Initial experimental results demonstrate that Factor-Q is able to learn as efficiently in dynamic as in static environments, unlike conventional methods. Further, in the specimen task,--obstacle avoidance is improved by over two orders--of magnitude compared with standard Qlearning.en
dc.language.isoeng
dc.relation.ispartofProcs TAROS
dc.rightsOpen
dc.titleImproving Learning for Embodied Agents in Dynamic--Environments by State Factorisationen
dc.contributor.institutionSchool of Computer Science
dc.description.statusPeer reviewed
dc.relation.schoolSchool of Computer Science
dcterms.dateAccepted2004
rioxxterms.typeJournal Article/Review
herts.preservation.rarelyaccessedtrue
herts.rights.accesstypeOpen


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record