Arcade

A collection of reinforcement learning projects.

View the Project on GitHub MGallow/Arcade

Back to Arcade.

Flappy Bird

Overview

We attempt to create a computer program that learns how to play Atari's popular Breakout game. This project utilizes reinforcement learning (deep Q-learning) and the Keras deep learning library. For background on CNN's, check out this excellent blog post by Andrej Karpathy. I would also recommend taking a look at this tutorial by Michael Nielsen for a general overview of neural networks. For deep Q-learning, checkout DeepMind's paper.

(source: https://yanpanlau.github.io/2016/07/10/FlappyBird-Keras.html)

Files:

Acknowledgements

Back to Arcade.