During my research, I read many papers and learned about many different reinforcement learning methods. I chose to implement a few of them in specific OpenAI Gym environments:
LunarLander-v2 Discrete LunarLander-v2 Continuous Retro Pong-v0
The reinforcement learning methods I chose to implement were dueling deep Q-learning, policy gradient learning and deep deterministic policy gradient learning. I also added keyboard code in order to facilitate human-in-the-loop learning and shared control schemes based on user-defined decision rules.
The branches of this repository will contain my implementations. I followed many tutorials for these implementations, so I obtained code or drew inspiration from many different sources ( I have pasted the links in either commented code or in the readme files).