This is the Generative Adversarial Network Demo written in Python using the Theano ML library. This is the code for the Generative Adversarial Network Episode of Fresh Machine Learning on Youtube. There is a default static gaussian distribution curve that the generator curve continously tries to mimic. Each timestep it gets better and better as it slowly gets better at fooling the disciminator network. Here's a cool live demo
- Python 2.7+ - (https://www.python.org/downloads/)
- matplotlib -
pip install matplotlib - theano -
pip install theano - scipy -
pip install scipy - foxhound (included)
Use pip to install any missing dependencies
If your dependencies are installed you can just run the code! A pyplot GUI should pop up and it will get better and better as it trains.
python demo.pyPull Requests are encouraged!!! This pyplot stuff can get wonky, so definitely make a PR if you see something that needs fixing.
Credit for this demo code Alec Radford and can be found here. I've merely moved some functions around and added a lot more documentation.