Repository in under construction

The algorithm learns to beat traditional data using pixel data from game environment(the game environment is in the repository itself.) (The game is traditional snake game with food appearing on random states.)

As one can see the model outperforms the environment getting almost full possible score.The model qalso learns to avoid different traps by the game.

1. David Silver Lecture
2. Sutton and Barto Book : Reinforcement Learning : An introduction
3. Actor-Critic Methods: A3C and A2C
4. Asynchronous Methods for Deep Reinforcement Learning
5. Continuous control with deep reinforcement learning
6. Playing Atari with Deep Reinforcement Learning

Repository is still building I am trying Actor Critic Models.The result will be uploaded soon.