Algorithm Selection on Continuous problems using Deep Learning.

Abstract

Selecting the most appropriate algorithm to use when tackling a black-box continuous optimization problem is a challenging task. with a wide range of optimization algorithms being benchmarked each year and easy access to black-box optimization functions, primarily due to the efforts of the COCO platform, an automated method for algorithm selection in single-objective black-box optimization problems has become necessary. In this issertation, we have chosen to employ DL technology in combination with ELA techniques to predict the optimal solver (SBS) for any given problem set. The performance of the proposed DL-based model is superior to a framework based on machine learning algorithms when applied to a continuous black-box optimization problem set, showcasing its effectiveness.