Strategic Features for General Games

Cameron Browne; Dennis Soemers; Eric Piette

Strategic Features for General Games

Cameron Browne, Dennis Soemers, Eric Piette

Advanced Computing Sciences

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceeding › Academic › peer-review

55 Downloads (Pure)

Abstract

This short paper describes an ongoing research project that requires the automated self-play learning and evaluation of a large number of board games in digital form.We describe the approach we are taking to determine relevant features, for biasing MCTS playouts for arbitrary games played on arbitrary
geometries. Benefits of our approach include efficient implementation, the potential to transfer learnt knowledge to new contexts, and the potential to explain strategic knowledge embedded in features in human-comprehensible terms.

Original language	English
Title of host publication	AAAI conference on Artificial Intelligence
Subtitle of host publication	Knowledge Extraction from Games (KEG'19)
Pages	71-75
Number of pages	5
Publication status	Published - 27 Jan 2019

Access to Document

Author's versionSubmitted manuscript, 143 KBLicence: CC BY-NC-ND

Cite this

@inproceedings{20c72bbf5e2a4a47b9c1d0b463b2be85,

title = "Strategic Features for General Games",

abstract = "This short paper describes an ongoing research project that requires the automated self-play learning and evaluation of a large number of board games in digital form.We describe the approach we are taking to determine relevant features, for biasing MCTS playouts for arbitrary games played on arbitrarygeometries. Benefits of our approach include efficient implementation, the potential to transfer learnt knowledge to new contexts, and the potential to explain strategic knowledge embedded in features in human-comprehensible terms.",

author = "Cameron Browne and Dennis Soemers and Eric Piette",

note = "Funding Information: This research is part of the European Research Council-funded Digital Ludeme Project (ERC Consolidator Grant #771292) being run by Cameron Browne. We would also like to acknowledge the RIKEN Institute{\textquoteright}s Advanced Intelligence Project (AIP), especially Kazuki Yoshizoe, for their generous support of prior research that led to this work. Publisher Copyright: {\textcopyright} 2019 CEUR Workshop Proceedings. All rights reserved.",

year = "2019",

month = jan,

day = "27",

language = "English",

pages = "71--75",

booktitle = "AAAI conference on Artificial Intelligence",

}

TY - GEN

T1 - Strategic Features for General Games

AU - Browne, Cameron

AU - Soemers, Dennis

AU - Piette, Eric

N1 - Funding Information: This research is part of the European Research Council-funded Digital Ludeme Project (ERC Consolidator Grant #771292) being run by Cameron Browne. We would also like to acknowledge the RIKEN Institute’s Advanced Intelligence Project (AIP), especially Kazuki Yoshizoe, for their generous support of prior research that led to this work. Publisher Copyright: © 2019 CEUR Workshop Proceedings. All rights reserved.

PY - 2019/1/27

Y1 - 2019/1/27

N2 - This short paper describes an ongoing research project that requires the automated self-play learning and evaluation of a large number of board games in digital form.We describe the approach we are taking to determine relevant features, for biasing MCTS playouts for arbitrary games played on arbitrarygeometries. Benefits of our approach include efficient implementation, the potential to transfer learnt knowledge to new contexts, and the potential to explain strategic knowledge embedded in features in human-comprehensible terms.

AB - This short paper describes an ongoing research project that requires the automated self-play learning and evaluation of a large number of board games in digital form.We describe the approach we are taking to determine relevant features, for biasing MCTS playouts for arbitrary games played on arbitrarygeometries. Benefits of our approach include efficient implementation, the potential to transfer learnt knowledge to new contexts, and the potential to explain strategic knowledge embedded in features in human-comprehensible terms.

M3 - Conference article in proceeding

SP - 71

EP - 75

BT - AAAI conference on Artificial Intelligence

ER -

Strategic Features for General Games

Abstract

Access to Document

Fingerprint

Cite this