Skip navigation
Skip navigation

Market-based reinforcement learning in partially observable worlds

Kwee, Ivo; Hutter, Marcus; Schmidhuber, Jürgen

Description

Unlike traditional reinforcement learning (RL), market-based RL is in principle applicable to worlds described by partially observable Markov Decision Processes (POMDPs), where an agent needs to learn short-term memories of relevant previous events in order to execute optimal actions. Most previous work, however, has focused on reactive settings (MDPs) instead of POMDPs. Here we reimplement a recent approach to market-based RL and for the first time evaluate it in a toy POMDP setting.

dc.contributor.authorKwee, Ivo
dc.contributor.authorHutter, Marcus
dc.contributor.authorSchmidhuber, Jürgen
dc.contributor.editorG. Dorffner
dc.contributor.editorH. Bischof
dc.contributor.editorK. Hornik
dc.coverage.spatialVienna, Austria
dc.date.accessioned2015-09-02T05:54:59Z
dc.date.available2015-09-02T05:54:59Z
dc.date.created21-25 August 2001
dc.identifier.isbn978-3-540-42486-4
dc.identifier.issn0302-9743
dc.identifier.urihttp://hdl.handle.net/1885/15097
dc.description.abstractUnlike traditional reinforcement learning (RL), market-based RL is in principle applicable to worlds described by partially observable Markov Decision Processes (POMDPs), where an agent needs to learn short-term memories of relevant previous events in order to execute optimal actions. Most previous work, however, has focused on reactive settings (MDPs) instead of POMDPs. Here we reimplement a recent approach to market-based RL and for the first time evaluate it in a toy POMDP setting.
dc.description.sponsorshipThis work was supported by SNF grants 21-55409.98 and 2000-61847.00
dc.publisherSpringer Verlag
dc.relation.ispartofArtificial Neural Networks - ICANN 2001: International Conference Vienna, Austria, August 21–25, 2001 Proceedings
dc.relation.ispartofseriesArtificial Neural Networks - ICANN 2001
dc.rights© Springer-Verlag Berlin Heidelberg 2001. http://www.sherpa.ac.uk/romeo/issn/0302-9743/..."Author's post-print on any open access repository after 12 months after publication" from SHERPA/RoMEO site (as at 2/09/15).
dc.sourceArtificial Neural Networks - ICANN 2001 Proceedings
dc.subjectHayek system
dc.subjectreinforcement learning
dc.subjectpartial observable environment
dc.titleMarket-based reinforcement learning in partially observable worlds
dc.typeConference paper
local.identifier.citationvolume2130
dc.date.issued2001
local.identifier.absfor080199 - Artificial Intelligence and Image Processing not elsewhere classified
local.identifier.ariespublicationu3700390xPUB177
local.publisher.urlhttp://link.springer.com/
local.type.statusAccepted Version
local.contributor.affiliationHutter, M., Research School of Computer Science, The Australian National University
local.bibliographicCitation.startpage865
local.bibliographicCitation.lastpage873
local.identifier.doi10.1007/3-540-44668-0_120
dc.date.updated2016-06-14T08:55:55Z
local.identifier.scopusID2-s2.0-84865410615
CollectionsANU Research Publications

Download

File Description SizeFormat Image
Kwee et al Market based Reinforcement Learning 2001.pdf134 kBAdobe PDFThumbnail


Items in Open Research are protected by copyright, with all rights reserved, unless otherwise indicated.

Updated:  23 August 2018/ Responsible Officer:  University Librarian/ Page Contact:  Library Systems & Web Coordinator