Generating General Levels using Markov Chains

Abstract

The use of machine learning techniques for content generation has recently emerged on the scene. Procedural Content Generation via Machine Learning is the generation of game content by models that have been trained on existing game content. The aim of this paper is to generate general video game levels using Markov chains. For this purpose, we created a random level generator that generates level dataset in order to train Markov models. The results show that Markov chains can generate playable levels for a large variety of games in the General Video Game Level Generation Framework. The generated levels are also evaluated using agent based testing.