Machine-Learning-Flappy-Bird

Why we included this project

This HTML5 project is one of the clearest, most compact demonstrations of neuroevolution you can find. It sets a genetic algorithm loose on a Flappy Bird clone: ten birds, each running its own tiny neural network, play the game at once, get scored on how far they fly, and the best networks recombine and mutate into the next generation. Since everything runs in the browser with Phaser and the Synaptic library, you can watch the flock improve round after round and then dig straight into the code. The accompanying tutorial breaks down the network architecture and each genetic operator step by step, which makes it a genuinely good on-ramp for anyone learning how evolutionary algorithms train neural controllers without labeled data. It is a teaching example rather than a production library, but as a visual introduction to the topic it is hard to beat.