. Tools used: Pygame(for developing the game interface), Python NEAT (Neuro Evolution of Augmenting Topologies)
Inputs
: Bird Y (Position of bird on y axis), Top Pipe (position of Top Pipe), Bottom Pipe (position of Bottom Pipe)Output
: Whether to make the bird jump or not jumpActivation Function
: TanHPopulation Size
: 10 Birds (can take any larger number, but taking larger size will overfit the model, as it's a very small game)Fitness Function
: The fitness of the Bird is dependent upon how long it plays the game (how far it can go)Maximum Generations
: 30, After running 30 Generations, if the Algorithm couldn't crack the game, then it means it failed(in our test case)
How the birds are trained in each generation? :