Applying LCS/XCS to the RTS Games Domain

Real-Time Strategy games (RTS) are representatives of the highest class of computational complexity in computer game genres. To cope with the high complexity of the state-action space of RTS game worlds, various Machine Learning algorithms are being used and researched extensively. In this article,...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Damijan Novak*, Domen Verber
Formato: article
Lenguaje:EN
Publicado: Faculty of Mechanical Engineering in Slavonski Brod, Faculty of Electrical Engineering in Osijek, Faculty of Civil Engineering in Osijek 2021
Materias:
AI
LCS
XCS
Acceso en línea:https://doaj.org/article/fc773c34b3884662a0a62469cd512946
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
Descripción
Sumario:Real-Time Strategy games (RTS) are representatives of the highest class of computational complexity in computer game genres. To cope with the high complexity of the state-action space of RTS game worlds, various Machine Learning algorithms are being used and researched extensively. In this article, we apply eXtended Classifier Systems (XCS) to the domain of RTS games. The XCS algorithm belongs to a Learning Classifier Systems (LCS) group known for their adaptability, generalisation, and scalability. We build the game agent named AIXCS. It uses a group of XCS algorithms, which generate a set of unit-actions used in the RTS game. The AIXCS operates without prior learning from the game runs and in tight timing constraints. The AIXCS was put to the test against other game agents in the micro RTS game environment, with positive results regarding successful game operation at runtime.