Two Gaussian Approaches to Black-Box Optomization

TL;DR

This paper explores two Gaussian process-based methods to enhance black-box optimization, specifically improving the covariance matrix adaptation evolution strategy (CMA-ES) through surrogate modeling techniques.

Contribution

It introduces two novel Gaussian approaches for surrogate modeling in black-box optimization, advancing the integration of Gaussian processes with CMA-ES.

Findings

Improved optimization efficiency with Gaussian surrogate models

Enhanced convergence rates demonstrated in benchmark tests

Potential for reduced function evaluations in complex problems

Abstract

Outline of several strategies for using Gaussian processes as surrogate models for the covariance matrix adaptation evolution strategy (CMA-ES).