Hands-On Neuroevolution with Python

Hypercube-based NEAT for visual discrimination

This chapter was the first of four chapters in which we discussed advanced neuroevolution methods. In this chapter, you learned about the indirect genome encoding scheme, which uses the Compositional Pattern Producing Network (CPPN) to aid with the encoding of large phenotype ANN topologies. The CPPN encoding scheme introduced by the NEAT extension is named HyperNEAT. This extension is built around the concept of the connectivity substrate that represents the phenotype ANN topology. At the same time, connections between nodes in the substrate are expressed as four-dimensional points within the hypercube. In the HyperNEAT method, the topology of the CPPN is the part that is evolving and guided ...