Hands-On Neuroevolution with Python

Hypercube-based NeuroEvolution of Augmenting Topologies

The methodology described in the previous section was invented by Kenneth O. Stanley and was called Hypercube-based NeuroEvolution of Augmenting Topologies (HyperNEAT). As its name suggests, it is an extension of the NEAT algorithm that we have already used in this book. The main difference between these two methods is that the HyperNEAT method uses an indirect encoding scheme based on the CPPN. During the evolution, the HyperNEAT method employs a NEAT algorithm to evolve a population of genomes that encode a topology of the connective CPPN. After that, each created CPPN can be used to establish the connectivity patterns within a specific phenotype ANN. Finally, the phenotype ANN can ...