Part-3 of the post of the navigation project of the DeepRL course by Udacity. We will cover two improvements to the base DQN implementation, namely Double DQN and Prioritized Experience Replay, as well as give some preliminary results of this improvements in the project environment. Finally we wrap up the post with some final remarks and ideas for future work.

Part-2 of the post of the navigation project of the DeepRL course by Udacity. We will cover the implementation of the DQN algorithm from Mnih et. al., discuss some gotchas during testing and tuning hyperparameters, and also give some the results of our implementation.

Part-1 of the post of the navigation project of the DeepRL course by Udacity. We will cover the environment setup to run our project submission, give an brief overview of RL and also an overview of the DQN algorithm from Mnih et. al.

This is a full post accompanying the submission of the navigation project of the DeepRL course by Udacity. We will cover how to solve the Banana Collector environment from Unity ml-agents using Deep Q-Learning. We will also cover two improvements for the base Deep Q-Learning algorithm, namely Double Q-learning and Prioritized Experience Replay.