MoveMentor

Kurzanzeige
dc.contributorMathys, Melanie
dc.contributorBarth, Can-Elian
dc.contributorGraber, Michael
dc.date.accessioned2026-10-14T13:11:55Z
dc.description.abstractModern day neural chess engines such as AlphaZero are trained via self-play to beat the most elite of human chess players, however, their decision making processes are completely opaque to us. In this project, we used CLIP as well as model distillation techniques to extract natural language explanations from highly performant Chess engines.
dc.identifier.urihttps://irf.fhnw.ch/handle/11645/58406
dc.subjectChess
dc.subjectCLIP
dc.subjectExplainable AI
dc.subjectModel distillation
dc.subjectReinforcement learning
dc.subject.ddc004 - Computer Wissenschaften, Internet
dc.subject.ddc790 - Freizeitgestaltung, Darstellende Kunst
dc.titleMoveMentor
dc.type00 - Projekt
dspace.entity.typeProject
fhnw.InventedHereYes
fhnw.Project.ContactPanos, Brandon
fhnw.Project.End2026-06-30
fhnw.Project.FinanceHasler Stiftung
fhnw.Project.ManagerPanos, Brandon
fhnw.Project.PartnersChessLine
fhnw.Project.Start2025-07-01
fhnw.Project.Stateabgeschlossen
fhnw.Project.Typeangewandte Forschung
fhnw.affiliation.hochschuleHochschule für Informatik FHNWde_CH
fhnw.affiliation.institutInstitut für Data Science
fhnw.sap.referenceI336-0051-2
relation.isContributorOfProjectc94f2103-6b86-4da9-ba33-c9427a598fc0
relation.isContributorOfProject1c879233-fda0-4956-93b2-e05c8f544c8b
relation.isContributorOfProject3693006b-11c7-4d96-9508-bd6f7c8b2301
relation.isContributorOfProject.latestForDiscoveryc94f2103-6b86-4da9-ba33-c9427a598fc0
relation.isProjectContactOfProject5cc45827-ef02-4fac-b0a2-7f3e223994d9
relation.isProjectContactOfProject.latestForDiscovery5cc45827-ef02-4fac-b0a2-7f3e223994d9
relation.isProjectManagerOfProject5cc45827-ef02-4fac-b0a2-7f3e223994d9
relation.isProjectManagerOfProject.latestForDiscovery5cc45827-ef02-4fac-b0a2-7f3e223994d9
Dateien

Lizenzbündel

Gerade angezeigt 1 - 1 von 1
Vorschaubild
Name:
license.txt
Größe:
2.66 KB
Format:
Item-specific license agreed upon to submission
Beschreibung:
Herunterladen
Sammlung
Hochschule für Informatik Projekte