Show simple item record

UCD: Upper Confidence Bound for Rooted Directed Acyclic Graphs

dc.contributor.authorMéhat, Jean
dc.contributor.authorCazenave, Tristan
HAL ID: 743184
dc.contributor.authorSaffidine, Abdallah
dc.date.accessioned2011-09-19T10:06:09Z
dc.date.available2011-09-19T10:06:09Z
dc.date.issued2010
dc.identifier.urihttps://basepub.dauphine.fr/handle/123456789/6978
dc.language.isoenen
dc.subjectDAGen
dc.subjectTranspositionsen
dc.subjectUCTen
dc.subjectMonte-Carlo Tree Searchen
dc.subject.ddc006.3en
dc.titleUCD: Upper Confidence Bound for Rooted Directed Acyclic Graphsen
dc.typeCommunication / Conférence
dc.contributor.editoruniversityotherLIASD Universit´e Paris 8 Saint-Denis France;France
dc.description.abstractenIn this paper we present a framework for testing various algorithms that deal with transpositions in Monte-Carlo Tree Search (MCTS). When using transpositions in MCTS, a Directed Acyclic Graph (DAG) is progressively developed instead of a tree. There are multiple ways to handle the exploration exploitation dilemma when dealing with transpositions. We propose parameterized ways to compute the mean of the child, the playouts of the parent and the playouts of the child. We test the resulting algorithms on LeftRight an abstract single player game and on Hex. For both games, original configurations of our algorithms improve on state of the art algorithms.en
dc.identifier.citationpages467-473en
dc.relation.ispartoftitleInternational Conference on Technologies and Applications of Artificial Intelligenceen
dc.relation.ispartofpublnameIEEEen
dc.relation.ispartofdate2010
dc.relation.isversionofdoihttp://dx.doi.org/10.1109/TAAI.2010.79
dc.description.sponsorshipprivateouien
dc.subject.ddclabelIntelligence artificielleen
dc.relation.ispartofisbn978-1-4244-8668-7en
dc.relation.conftitleTAAI 2010en
dc.relation.confdate2010-11
dc.relation.confcityHsinchu Cityen
dc.relation.confcountryTaïwanen


Files in this item

FilesSizeFormatView

There are no files associated with this item.

This item appears in the following Collection(s)

Show simple item record