Uploaded February 2018 | Updated November 2026, 3 hours ago
SorI/O is a program made of neural networks and genetic algorithms that learns how to play Megadrive games such as Streets of Rage.
Source Code: github.com/gsaurus/SorIO
After learning to play the first scene of stage 1 using Axel in normal mode (until the entrance of the bar), it was commanded to play as Blaze on easy for the entire game.
The algorithm never played as Blaze before, and never went past first scene of stage 1. Surprisingly it made it all the way to the final boss, Mr. X.
Based on MarI/O: youtube.com/watch?v=qv6UVOQ0F44
"NEAT" paper in which the algorithm is based on:
http://nn.cs.utexas.edu/downloads/papers/stanley.ec02.pdf
SorI/O is a program made of neural networks and genetic algorithms that learns how to play Megadrive games such as Streets of Rage.
Source Code: github.com/gsaurus/SorIO
After learning to play the first scene of stage 1 using Axel in normal mode (until the entrance of the bar), it was commanded to play as Blaze on easy for the entire game.
The algorithm never played as Blaze before, and never went past first scene of stage 1. Surprisingly it made it all the way to the final boss, Mr. X.
Based on MarI/O: youtube.com/watch?v=qv6UVOQ0F44
"NEAT" paper in which the algorithm is based on:
http://nn.cs.utexas.edu/downloads/papers/stanley.ec02.pdf










