Adversarial Random Forest Classifier for Automated Game Design

TL;DR

This paper explores an adversarial approach to learning a fitness function for autonomous game design using a Random Forest classifier, aiming to reduce reliance on human-authored knowledge, but the experiment did not fully meet expectations.

Contribution

It introduces an adversarial learning method for automatic game design, attempting to mimic human-like fitness functions without extensive human input.

Findings

The adversarial approach provided insights into fitness function learning.

Experimental results did not fully meet initial expectations.

Analysis offers guidance for future autonomous game design research.

Abstract

Autonomous game design, generating games algorithmically, has been a longtime goal within the technical games research field. However, existing autonomous game design systems have relied in large part on human-authoring for game design knowledge, such as fitness functions in search-based methods. In this paper, we describe an experiment to attempt to learn a human-like fitness function for autonomous game design in an adversarial manner. While our experimental work did not meet our expectations, we present an analysis of our system and results that we hope will be informative to future autonomous game design research.