Tanner Laborde
This is an example of a genetic algorithm constructing a neural network to beat Super Mario Bros. This is the fastest run after around 200 generations. The fitness function is defined by speed. In the future I may modify the algorithm to factor in game score. If you are interested in learning how to create your own machine learning algorithm for games leave a comment below. If their is enough interest I will create a tutorial.
This is based off of the work done by Sethbling with MarI/O. Look up his video if you are interested.
AI Landing on the moon:
https://www.youtube.com/watch?v=_pF1Un5zvp4
Source
Nice video. Im interested in learning to make machine learning algorithms.
Btw what is the song?
It can't go left.
You should credit the creator and his video with a link.
Did you only train it on the first level? How does it perform on other levels? Just wondering if it's overfitting to the first level.
Great work Tanner, if you could upload more videos about that, would be great
can you share the code?
My brain netwaorks r better.
Just deign the enemy that breaks that jumping pattern and skynet lose
It would be interesting to add a new feature in order to get maximum points in a level.
????????????????????????????????????????????????????????????????????????☕?????????? which food would you eat and which drink would you drink