TY - JOUR

T1 - Exception-Tolerant Hierarchical Knowledge Bases for Forward Model Learning

AU - Apeldoorn, Daan

AU - Dockhorn, Alexander

N1 - Funding Information: Manuscript received January 20, 2020; revised April 25, 2020; accepted June 19, 2020. Date of publication July 8, 2020; date of current version September 15, 2021. This work was supported by IMBEI. (Corresponding author: Alexander Dockhorn.) Daan Apeldoorn is with the Institute for Medical Biostatistics, Epidemiology and Informatics (IMBEI) in the Medical Informatics Department, University Medical Centre, Johannes Gutenberg University Mainz, 55122 Mainz, Germany (e-mail: daan.apeldoorn@uni-mainz.de).

PY - 2021/9

Y1 - 2021/9

N2 - This article provides an overview of the recently proposed forward model approximation framework for learning games of the general video game artificial intelligence (GVGAI) framework. In contrast to other general game-playing algorithms, the proposed agent model does not need a full description of the game but can learn the game's rules by observing game state transitions. Based on hierarchical knowledge bases, the forward model can be learned and revised during game-play, improving the accuracy of the agent's state predictions over time. This allows the application of simulation-based search algorithms and belief revision techniques to previously unknown settings. We show that the proposed framework is able to quickly learn a model for dynamic environments in the context of the GVGAI framework.

AB - This article provides an overview of the recently proposed forward model approximation framework for learning games of the general video game artificial intelligence (GVGAI) framework. In contrast to other general game-playing algorithms, the proposed agent model does not need a full description of the game but can learn the game's rules by observing game state transitions. Based on hierarchical knowledge bases, the forward model can be learned and revised during game-play, improving the accuracy of the agent's state predictions over time. This allows the application of simulation-based search algorithms and belief revision techniques to previously unknown settings. We show that the proposed framework is able to quickly learn a model for dynamic environments in the context of the GVGAI framework.

KW - Belief revision

KW - Forward Model Approximation (FMA)

KW - general video games

KW - Hierarchical Knowledge Bases (HKBs)

KW - Monte Carlo Tree Search (MCTS)

UR - http://www.scopus.com/inward/record.url?scp=85115247260&partnerID=8YFLogxK

U2 - 10.1109/TG.2020.3008002

DO - 10.1109/TG.2020.3008002

M3 - Article

AN - SCOPUS:85115247260

VL - 13

SP - 249

EP - 262

JO - IEEE Transactions on Games

JF - IEEE Transactions on Games

SN - 2475-1502

IS - 3

ER -