Hands-On Neuroevolution with Python

PyTorch NEAT

This library is built around the NEAT-Python library. It provides easy integration for artifacts that have been produced by the NEAT-Python library with the PyTorch platform. As a result, it becomes possible to convert the NEAT genome into a phenotype ANN, which is based on the PyTorch implementation of recurrent neural networks. Also, it allows us to represent Compositional Pattern Producing Networks (CPPNs) as PyTorch structures, which are the primary building blocks of the HyperNEAT method. The main advantage of integration with PyTorch is that it allows us to utilize GPUs for computing, potentially accelerating the evolutionary process due to the increased rate of evaluation of the genomes of organisms in the evolving population. ...