Hands-On Neuroevolution with Python

Hyperparameter selection

The MultiNEAT library uses the Parameters Python class to hold all the required hyperparameters. To set the appropriate values of the hyperparameters, we define the create_hyperparameters function in the experiment runner Python script. Here, we describe the essential hyperparameters that have a significant impact on the HyperNEAT algorithm performance in this experiment:

1. The create_hyperparameters function begins by creating a Parameters object to hold the HyperNEAT parameters:

    params = NEAT.Parameters()

2. We decided to start with a medium-sized population of genomes to keep the computations fast. At the same time, we want to maintain a sufficient number of organisms in the population for evolutionary diversity. ...