Theano implementation of one step Q-learning: Asynchronous Methods for Deep Reinforcement Learning
Theano and Arcade-Learning-Environement
-
Theano: website pip install theano
-
Arcade learning environment :
git clone https://github.com/mgbellemare/Arcade-Learning-Environment
cd Arcade-Learning-Environment
mkdir buid
cd build
cmake -DUSE_SDL=OFF -DUSE_RLGLUE=OFF -DBUILD_EXAMPLES=OFF ..
make
pip install ..
python trainAgent.py breakout.bin