Info
This post was imported from a personal note. It may contain inside jokes, streams of consciousness, errors, and other nonsense.
Implementation of NEAT https://github.com/janhohenheim/Hippocrates
Implementation of rtNEAT https://github.com/qwfy/rtNEAT
Can probably learn a lot about C++ from these examples, too.
NEAT is NeuroEvolution of Augmenting Topologies. As in evolving the structure of the network along with the weights of the network. NeuroEvolution (NE) initializes a bunch of agents using networks with random weights. Then it lets them act for a while in the simulation. Then it takes the best of them and maybe breeds them and puts them back in the simulation.
So to start with seems like it’d be alright for me to use just a few nodes with random weights and let them run around and then iterate like that.