Related work

J. Biles developed since 1994 GenJam, a genetic jammer to play jam sessions with. It represents the state of the art in the field and produced important studies in music representation with genetic algorithms, which have been also used to form the basis of this work. Also, the idea of an interactive system that lets the user evaluate the produced patterns comes from his work. Combined with a good rendering engine for MIDI events this system passed the Turing test against its own author, making a human think its flute line was less mechanical than Biles’ saxophone. However, as said before, this work is aimed at a professional interactive music system, and is not suited for game play.

Erin Jonathan Hastings, Ratan K. Guha have shown with their Galactic Arms Race video game that it is possible to evolve aesthetically pleasing and working weapons for a game using genetic algorithms to balance the weights of a neural network that produces the features of each weapon. Again in this game, the evaluation of the population is left to the user, showing a small set of possible weapons that the player can use or discard, depending on his own interest. Albeit this work is intended for a different type of video games, its results are encouraging and their ideas on population selection have been adapted.