NN-Racer is a recreation of a project I worked on a few years ago while learning about neural networks and genetic algorithms. It’s a simple implementation, written from scratch, of a neural network that evolves by mutating the weights and biases of the best-performing cars in each generation.
Each weight or bias has a 1 in 100 chance of mutation, with the mutation range being from -0.5 to 0.5 times its original value.
You can try it here: NN-Racer
[Stack: Unity, C#, AI, NN, GA]
