Hands-On Neuroevolution with Python

The evaluate_novelty_score function

This function provides a mechanism to evaluate the novelty score of the novelty item against all items already collected in the novelty archive and all the novelty items discovered in the current population. We calculate the novelty score as the average distance to the k=15 nearest neighbors by following these steps:

1. We need to collect the distances from the provided novelty item to all items in the novelty archive:

        distances = []        for n in self.novel_items:            if n.genomeId != item.genomeId:                distances.append(self.novelty_metric(n, item))            else:                print("Novelty Item is already in archive: %d" %                        n.genomeId)

2. After that, we add the distances from the provided novelty item to all items in the current population: ...