Towards Game Design via Creative Machine Learning (GDCML)

Abstract

In recent years, machine learning (ML) systems have been increasingly applied for performing creative tasks. Such creative ML approaches have seen wide use in the domains of visual art and music for applications such as image and music generation and style transfer. However, similar creative ML techniques have not been as widely adopted in the domain of game design despite the emergence of ML-based methods for generating game content. In this paper, we argue for leveraging and repurposing such creative techniques for designing content for games, referring to these as approaches for Game Design via Creative ML (GDCML). We highlight existing systems that enable GDCML and illustrate how creative ML can inform new systems via example applications and a proposed system.

Generating and Blending Game Levels via Quality-Diversity in the Latent Space of a Variational Autoencoder

“…Finally, in generating levels blended across different games, our work is situated amidst a recent line of PCGML research focusing on more creative applications of ML for game design [19,33]. Such techniques touch upon concepts of combinational creativity [2] and have included domain transfer [39,42], automated game generation [18] and game blending [13] which refers to generating new games by combining properties such as levels and/or mechanics of existing games.…”

Adversarial Random Forest Classifier for Automated Game Design

“…Some researchers have demonstrated the ability to learn game mechanics from human-authored games and apply this to the generation of novel mechanics and levels [5,7,23]. There have also been proposals for other complete ML-based automated game design systems [12,17], however, only components of these systems have been implemented at this point [23]. Thus far, all of these approaches focus on platformer games, where our system focuses on Video Game Description Language (VGDL) games, which represent a number of game genres.…”

“…Currently the space of PCGML game generation is under-explored. Researchers have proposed PCGML game generators [12,17], but only a few systems have been built that can generate game mechanics, and these generally focus on specific mechanics or genres [7,23]. This lack of implementation is due, in part, to the inherent complexity of games and the rules that define them.…”

Game Design as an Autonomous Research Subject

“…Formal methods never completely went away, but interest waned. Since then, method proposals on the part of academic researchers have been more likely to variously pertain to specific aspects of games such as for instance motivation [20], level design [21] in particular genres, or sound [22] as opposed to a general method, or to be integrated with other disciplinary concerns, such as learning [23], analytics [24], or artificial intelligence [25]. There has been some interest in general methods on the part of the games industry, such as Ubisoft's rational design [26], and academia has occasionally been returning to the idea of a general method, but game methodology remains fragmentary.…”