This is a practice project of reinforcement learning. A Deep Q-Network (DQN) model was trained with PARL framework to play the game Flappy Bird. Modification of the parameter settings or longer training time may be necessary to obtain a better score.

python train.py

python test.py

1. paddlepaddle: https://github.com/PaddlePaddle/Paddle
2. PARL: https://github.com/PaddlePaddle/PARL
3. PyGame-Learning-Environment: https://github.com/ntasfi/PyGame-Learning-Environment

1. https://github.com/PaddlePaddle/PARL/tree/develop/examples
2. https://github.com/ninetailskim/FlappyPaddle
3. https://github.com/Ryan906k9/fly_bird_RL