Generating Tailored Advice in Video Games through Play-Style Identification and Player Modelling

Abstract

Recent advances in fields such as reinforcement learning have enabled the development of systems that are able to achieve super-human performance on a number of domains, specifically in complex games such as Go and StarCraft. Based on these successes, it is reasonable to ask if these learned behaviours could be utilised to improve the performance of humans on the same tasks. However, the types of models used in these systems are typically not easily understandable, and can not be directly used to improve the performance of a human. My research looks to address these difficulties by developing a system that can provide advice tailored to a player's style in a video game setting. This system would be particularly useful for improving tutorial systems that quickly become redundant lacking any personalization. I have already developed an unsupervised approach to identifying different play-styles present in a video game. I look to use this knowledge to train agents which can demonstrate optimal behaviour for each style. It is hoped that this information can be utilised to generate useful advice based on the differences between the current performance of a player and the corresponding expert agent in their style.