Reinforcement-Learning

My reinforcement library (for the "Firefly" task)

As of now it contains implementation of Double DQN which runs on a simplified version of the task (with discrete actions). I'm planning to implement Prioritized Experience Replay soon.

The actual nature of the firefly task has continuous action space and hence I plan to implement Policy gradient and actor-critic methods.

It also has a new OpenAI Gym environments.

Eventually, we need to map the model parameters to neural data.