Hands-On Neuroevolution with Python

NEAT encoding scheme

The genetic encoding scheme of NEAT is designed to allow easy matching between corresponding genes during the mating process when a crossover operator is applied to the two parent genomes. The NEAT genome is a linear representation of the connectivity pattern of the encoded neural network, as shown in the following NEAT genome scheme:

The NEAT genome scheme

Each genome is represented as a list of connection genes that encode connections between the nodes of the neural network. Also, there are node genes that encode information about network nodes, such as the node identifier, node type, and type of activation function. ...