A reinforcement learning agent for OpenAI Gym.
- Install Python v2.7+
- (Optional) Invoke a virtual environment via
virtualenv
- Install dependencies by running
pip install -r requirements.txt
- Run
./run
to reproduce experiments (graphs may differ due to randomness)
NOTE: If Box2D is causing errors, you may need to build from source. Run
git submodule init && git submodule update
and./build_pybox2d
.
- Bill Learning's YouTube video, https://www.youtube.com/watch?v=Lv_VDz1RhWY
- Fiszel, Reinforcement Learning and DQN, https://rubenfiszel.github.io/posts/rl4j/2016-08-24-Reinforcement-Learning-and-DQN.html
- Hasselt et al, Deep Reinforcement Learning with Double Q-Learning (2015), https://arxiv.org/pdf/1509.06461.pdf
- Matiisen, Demystifying Deep Reinforcement Learning (2015), https://www.intelnervana.com/demystifying-deep-reinforcement-learning/
- Minh et al, Playing Atari with Deep Reinforcement Learning (2015), https://www.cs.toronto.edu/~vmnih/docs/dqn.pdf