An Unsupervised Video Game Playstyle Metric via State Discretization

TL;DR

This paper introduces an unsupervised, observation-based metric for analyzing video game playstyles by discretizing game states, enabling comparison across different agents and games without prior feature engineering.

Contribution

It presents a novel method for learning discrete state representations from game observations to measure and compare playstyles without predefined features.

Findings

High accuracy in playstyle discrimination across multiple games

Effective for different agent types including humans and AI bots

Applicable to various game platforms like TORCS, RGSK, and Atari

Abstract

On playing video games, different players usually have their own playstyles. Recently, there have been great improvements for the video game AIs on the playing strength. However, past researches for analyzing the behaviors of players still used heuristic rules or the behavior features with the game-environment support, thus being exhausted for the developers to define the features of discriminating various playstyles. In this paper, we propose the first metric for video game playstyles directly from the game observations and actions, without any prior specification on the playstyle in the target game. Our proposed method is built upon a novel scheme of learning discrete representations that can map game observations into latent discrete states, such that playstyles can be exhibited from these discrete states. Namely, we measure the playstyle distance based on game observations aligned to the same states. We demonstrate high playstyle accuracy of our metric in experiments on some video game platforms, including TORCS, RGSK, and seven Atari games, and for different agents including rule-based AI bots, learning-based AI bots, and human players.