A Machine-Learned Framework for Automatic Content Generation, Evaluation, and Critique

Summerville Adam J.
2016 DiGRA/FDG '16 - Abstract Proceedings of the First International Joint Conference of DiGRA and FDG

The majority of procedural level research has relied on human authored rules and heuristics. I intend to develop an end-to-end system capable of examining play, understanding levels and generating new content of similar style and playability, and finally offering analysis and critique of levels. The current roadmap utilizes computer vision, causal modeling, and neural network systems. The system should allow a human to step in at any point and make whatever changes they wish and get all downstream benefits.


Super Mario as a String: Platformer Level Generation Via LSTMs

Summerville Adam J. Mateas Michael
2016 DiGRA/FDG '16 - Proceedings of the First International Joint Conference of DiGRA and FDG

The procedural generation of video game levels has existed for at least 30 years, but only recently have machine learning approaches been used to generate levels without specifying the rules for generation. A number of these have looked at platformer levels as a sequence of characters and performed generation using Markov chains. In this paper we examine the use of Long Short-Term Memory recurrent neural networks (LSTMs) for the purpose of generating levels trained from a corpus of Super Mario Bros. levels. We analyze a number of different data representations and how the generated levels fit into the space of human authored Super Mario Bros. levels.