Hands-On Neuroevolution with Python

The experiment runner function

The run_experiment function runs the experiment using the provided hyperparameters and an initialized test environment to evaluate the discovered detector ANNs against the possible retina configurations. The function implementation has the following significant parts:

1. First is the initialization of the population of the initial CPPN genomes:

    seed = 1569777981    # Create substrate    substrate = create_substrate()    # Create CPPN genome and population    g = NEAT.Genome(0,             substrate.GetMinCPPNInputs(),             2, # hidden units             substrate.GetMinCPPNOutputs(),             False,             NEAT.ActivationFunction.TANH,             NEAT.ActivationFunction.SIGNED_GAUSS, # hidden              1, # hidden layers seed             params,              1) # one hidden layer pop = NEAT.Population(g, params, ...