GameDecisionMaking

Final Project

This project trains an agent to play the flappy bird by using various reinforcement learning algorithms. The trained model will be saved in directory saved_networks.

The main executable files:

• deep_q_network.py is implementation of DQN
• deep_q_network_l.py is implementation of DQN without the target network
• actor_critic_network.py is the implementation of actor_critic
• evaluation_network.py is for the performance evaluation of DQN and actor_critic
• model_free.py is implementation of SARSA, Q-learning and improved Q-learning
• simulator.py is implementation of playing the game to train SARSA, Q-learning and improved Q-learning model
• evaluation_model_free.py is for tje performance evaluation of SARSA, Q-learning and improved Q-learning model
• visulization.py is implementation of plot