Hands-On Neuroevolution with Python

The population fitness evaluation function

The fitness evaluation function is a callback function that is registered with the NEAT-Python library, allowing this library to run an evaluation of population genomes against specific conditions of a particular task that needs to be solved. We implement this function to evaluate each genome in the current population using the maze-solving task, and to use the obtained novelty score as a genome fitness value.

The NEAT-Python library doesn't allow us to send any signals about task completion from the callback function other than by specifying the specific fitness score value of the winner genome. This fitness value must be higher than the fitness threshold in the NEAT-Python hyperparameter configuration. ...