Hands-On Neuroevolution with Python

Fitness evaluation

The genome's fitness evaluation is a significant part of any neuroevolution algorithm, including the HyperNEAT method. As you've seen, the main experiment loop invokes the eval_genomes function to evaluate the fitness of all genomes within a population for each generation. Here, we consider the implementation details of the fitness evaluation routines, which consists of two main functions:

• The eval_genomes function evaluates all genomes in the population:

def eval_genomes(genomes, substrate, vd_environment, generation):    best_genome = None    max_fitness = 0    distances = []    for genome in genomes:        fitness, dist = eval_individual(genome, substrate,                                         vd_environment)        genome.SetFitness(fitness)        distances.append(dist) if fitness ...