Bootstrapping Conditional GANs for Video Game Level Generation

TL;DR

This paper introduces CESAGAN, a novel GAN architecture with a bootstrapping training method that improves the generation of diverse, playable, and aesthetically appealing video game levels with fewer duplicates.

Contribution

The paper proposes CESAGAN, a conditional self-attention GAN with a bootstrapping training procedure for improved game level generation.

Findings

Generates more playable levels with fewer duplicates.

Models non-local dependencies between game objects.

Reduces training data requirements through bootstrapping.

Abstract

Generative Adversarial Networks (GANs) have shown im-pressive results for image generation. However, GANs facechallenges in generating contents with certain types of con-straints, such as game levels. Specifically, it is difficult togenerate levels that have aesthetic appeal and are playable atthe same time. Additionally, because training data usually islimited, it is challenging to generate unique levels with cur-rent GANs. In this paper, we propose a new GAN architec-ture namedConditional Embedding Self-Attention Genera-tive Adversarial Network(CESAGAN) and a new bootstrap-ping training procedure. The CESAGAN is a modification ofthe self-attention GAN that incorporates an embedding fea-ture vector input to condition the training of the discriminatorand generator. This allows the network to model non-localdependency between game objects, and to count objects. Ad-ditionally, to reduce the number of levels necessary to trainthe GAN, we propose a bootstrapping mechanism in whichplayable generated levels are added to the training set. Theresults demonstrate that the new approach does not only gen-erate a larger number of levels that are playable but also gen-erates fewer duplicate levels compared to a standard GAN.