Training Atari Games using Deep Reinforcement Learning

Overview

Deep Q-Network (DQN) is the first deep reinforcement learning method proposed by DeepMind.

Here we will train a Tensorflow DQN implementation based on DeepMind's DQN for playing Atari games.

Either use Virtual Environment through venv or anaconda:

To run, type the following into the terminal

$ python main.py --game <env_name>

However, if you are on an Anaconda virtual environment no need to type <env_name> just type

$ python main.py --game SpaceInvaders-v0 --display true

and instead of SpaceInvaders-v0 just type any other game name.

Thank you Kee Hyun Won for inspiring this code. I just adjusted the code for my environment and for the new tensorflow deprecated variables.

Name		Name	Last commit message	Last commit date
Latest commit History 23 Commits
README.md		README.md
buff.py		buff.py
convnet.py		convnet.py
dqn.py		dqn.py
environment.py		environment.py
main.py		main.py
memory.py		memory.py