NN + RL for Flappy Birds
An agent trained to play Flappy Birds using Q learning and Neural Networks. Our project for the Neural Networks course in 2019.
The learning environment used is from PyGame-Learning-Environment.
Results of our best agent ( vertical -> score, horizontal -> number of games played):
Two hidden layers, with leaky relu and tanh as activation functions and Adadelta as optimizer.