Reinforcement-learning

Implement SR and Replay with DNN

DQN

Double_QDN_TEST_v2 :

Implement Simple and Double DQN on gridworld and catpole

DQN_ONE_HOT :

Upgrade Double DQN on discrete and continue space states with the same NN

LSFNN

Linear Successor Feature Neural Network Create a latent space allowing faster transfer learning on reward and transition.