DRL (Deep Reinforcement Learning Agent) is an implementation of the deep reinforcement learning algorithm, described in Playing Atari with Deep Reinforcement Learning paper by DeepMind. The original implementation of the DRL project was in theano and was based on TNNF - Tiny Neural Net Framework which used theano library for GPU computing.

The Tiny Neural Net Framework was revised to work on TensorFlow and hence this DRL project works on TensorFlow instead of theano. DRL is currently tested on Ubuntu 14.04 on GPU.

Games: Breakout (video) and Space Invaders

To install DRL on Ubuntu:

sudo apt-get install python-pil python-numpy python-scipy python-dev python-pip python-nose g++ libopenblas-dev git libsdl1.2-dev libsdl-image1.2-dev libsdl-gfx1.2-dev python-matplotlib libyaml-dev
sudo pip install -U numpy
sudo pip install -U pillow==2.7.0

To install TensorFlow, please follow the instructions on the TensorFlow webpage.
**Note: DRL requires CUDA and please follow the instructions to install CUDA through TensorFlow page.
Then navigate to DRL-TensorFlow/emulator directory and run:

cd DRL
bash ./install.sh

To launch the training, run the following:

python main_exp.py