CMA-ES black box optimization with Gaussian Process surrogate models

Datum konání: 05.04.2019
Přednášející: Lukáš Bajer
Odpovědná osoba: Kotera

The talk will focus on connection of Gaussian process, statistical/regression models, and Covariance Matrix Adaptation Evolutionary Strategy (CMA-ES), a black-box optimization algorithm. Lukas will discuss how the Gaussian processes can be used as surrogate models for the black-box continuous optimization, focusing on the benefits of employing the Gaussian process uncertainty prediction, especially during the selection of points for the evaluation with a surrogate model. The talk will conclude with experimental results of author's implementation of such combination, the DTS-CMA-ES algorithm.