TensorBird

Intro

use deep-q-learning algorithm to train agent to play flappy -bird game.
coding by python3, used frameworks:

• tensorflow
• opencv
• pygame

How to run

cd into project directory, type command in terminal:

python dqn.py

How to re-training

edit dqn.py:
comment code:

kEpsilonInit = 0.0001; 

rewrite it:

kEpsilonInit = 0.1000; 

re-run to start training.

Ref

[1] Playing FlappyBird with Deep Reinforcement Learning
[2] Github: DeepLearningFlappyBird