The Bayesian Optimization algorithm is a method used in a lot of domains, including machine leaning for finding hyperparameters. It allows to compute an approximation of a black-box function which is expensive to evaluate.
The implementation is based off the following paper: https://export.arxiv.org/pdf/1807.02811
The Gaussian kernel is used as the interpolation kernel, and the Expected Improvement function as the acquisition function.