Hands-On Genetic Algorithms with Python

Hyperparameter representation

In Chapter 6, Optimizing Continuous Functions, we used genetic algorithms to optimize the functions of real-valued parameters. These parameters were represented as a list of float numbers, like so:

[1.23, 7.2134, -25.309]

Consequently, the genetic operators we used were specialized for handling lists of floating-point numbers.

To adapt this approach so that it can tune the hyperparameters, we are going to represent each hyperparameter as a floating-point number, regardless of its actual type. To make this work, we need to find a way to transform each parameter into a floating-point number, and back from a floating-point number to its original representation. We will implement these transformations as follows: ...