Applies a Deep Q Network to play a game of Pong.
The weights determined from training are stored in weights.hdf5
.
The implementation of Pong is a modified version of https://github.com/MikeChunko/pong.
To run and train, make sure you have the keras, numpy, and pygame libraries installed, clone the repo, and run python pong.py
.
Run python pong.py --help
for a complete list of options.