Exercises on deep reinforcement learning using Tensorflow

PER: Prioritized Experience Replay

DQN: Deep Q-network

• Plain Q-learning to solve Frozenlake environment
• DQN for Cartpole
• DQN with PER
• Double DQN (DDQN) for Breakout (an Atari 2600 game)
• DDQN with PER
• Dueling DQN with PER
• A2C
• A3C

Learning aim: Getting familiarized in basic concepts of RL in order to apply them to my research involving medical imaging.

Requirements: I used Tensorflow (1.9) and Gym (0.9.6).