Automated Game Design via Conceptual Expansion

摘要: Automated game design has remained a key challenge within the field of Game AI. In this paper, we introduce method for recombining existing games to create new through process called conceptual expansion. Prior automated approaches have relied on hand-authored or crowd-sourced knowledge, which limits scope and applications such systems. Our approach instead relies machine learning learn approximate representations games. recombines knowledge from these learned via We evaluate by demonstrating ability system recreate To best our represents first learning-based system.