Neural Network Evolution Playground with Backprop NEAT

Each neuron and connection in the entire population is unique, and assigned a unique integer label. So each network is simply just a list of connections, along with the weight for those connections. Note that two different networks can have a similar connection, but the weights for each network will generally be different. Because each neuron and each connection is globally unique, it becomes possible to merge two different networks, to produce a new network.