Super Mario as a String: Platformer Level Generation Via LSTMs

TL;DR

This paper explores using LSTM neural networks to generate Super Mario Brothers levels by treating them as character sequences, analyzing different data representations and how well generated levels resemble human-designed ones.

Contribution

It introduces a novel application of LSTMs for platformer level generation and compares various data representations for improved generation quality.

Findings

LSTM models can generate playable Super Mario levels.

Different data representations affect the quality of generated levels.

Generated levels can resemble human-designed levels in structure.

Abstract

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 Brothers levels. We analyze a number of different data representations and how the generated levels fit into the space of human authored Super Mario Brothers levels.