Thumbnail

Improved Policy Networks for Computer Go

Cazenave, Tristan (2017), Improved Policy Networks for Computer Go, in Winands, Mark H.M.; van den Herik, H. Jaap; Kosters, Walter A., Advances in Computer Games, 15th International Conferences, ACG 2017, Revised Selected Papers, Springer : Berlin Heidelberg, p. 90-100. 10.1007/978-3-319-71649-7_8

View/Open
Type
Communication / Conférence
Date
2017
Conference title
15th International Conferences (ACG 2017)
Conference date
2017-07
Conference city
Leiden
Conference country
Netherlands
Book title
Advances in Computer Games, 15th International Conferences, ACG 2017, Revised Selected Papers
Book author
Winands, Mark H.M.; van den Herik, H. Jaap; Kosters, Walter A.
Publisher
Springer
Published in
Berlin Heidelberg
ISBN
978-3-319-71648-0
Number of pages
235
Pages
90-100
Publication identifier
Metadata
Show full item record
Author(s)
Cazenave, Tristan
Laboratoire d'analyse et modélisation de systèmes pour l'aide à la décision [LAMSADE]
Abstract (EN)
Golois uses residual policy networks to play Go. Two improvements to these residual policy networks are proposed and tested. The first one is to use three output planes. The second one is to add Spatial Batch Normalization.
Subjects / Keywords
distributed computing; combinatorial game; Go game