Self play with parameter sharing in n-player mixed competitive-cooperative games

Simple item page
dc.contributor.authorSkaltsis, George Marios
dc.contributor.authorShin, Hyosang
dc.contributor.authorTsourdos, Antonios
dc.date.accessioned2022-02-16T12:22:14Z
dc.date.available2022-02-16T12:22:14Z
dc.date.issued2021-12-29
dc.description.abstractWe introduce some parameter sharing multi-agent reinforcement learning schemes, combined with self-play for n-players mixed competitive-cooperative games. Except for the pure self-play scheme, an another one using the best policy, outperform the pure parameter sharing baseline, leading to better exploration of the state space and protecting from the performance deterioration observed with the baseline.en_UK
dc.identifier.citationSkaltsis GM, Shin H-S, Tsourdos A. (2021) Self play with parameter sharing in n-player mixed competitive-cooperative games. In: AIAA SciTech 2022 Forum, 3-7 January 2022, San Diego, CA and Virtual Eventen_UK
dc.identifier.urihttps://doi.org/10.2514/6.2022-2498
dc.identifier.urihttps://dspace.lib.cranfield.ac.uk/handle/1826/17576
dc.language.isoenen_UK
dc.publisherAIAAen_UK
dc.rightsAttribution-NonCommercial 4.0 International*
dc.rights.urihttp://creativecommons.org/licenses/by-nc/4.0/*
dc.subjectReinforcement Learningen_UK
dc.subjectMarkov Decision Processen_UK
dc.subjectStatistical Distributionsen_UK
dc.subjectOptimization Algorithmen_UK
dc.subjectMulti Agent Systemen_UK
dc.subjectComputer Systemsen_UK
dc.subjectValue Functionen_UK
dc.subjectNeural Networksen_UK
dc.subjectMathematical Modelsen_UK
dc.subjectGreedy Algorithmen_UK
dc.titleSelf play with parameter sharing in n-player mixed competitive-cooperative gamesen_UK
dc.typeConference paperen_UK

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Self_play_with_parameter_sharing-2021.pdf
Size:
1.67 MB
Format:
Adobe Portable Document Format
Description:
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.63 KB
Format:
Item-specific license agreed upon to submission
Description:
Download

Collections

Staff publications (SATM)