How does Selecting a Benchmark Function Suite Influence the Estimation of an Algorithm's Quality?
Iztok Fister, Suash Deb, Iztok Fister, Iztok Fister · 2019
This paper is focused on answering the question how the selection of a testbed on which the newly proposed algorithms are evaluated influence t he estimation of an algorithm's quality. New algorithms are usually tested on well-known benchmark function suites, where the goal is to achieve the best results of the algorithm in the shortest time. A lot of questions have arisen when looking for the most suitable testbed, for instance, which benchmark to take, and which version of it is the most representative for determining the best algorithms. In this study, the newly proposed algorithms introducing the coalition game concept for solving global optimization were tested by solving two different benchmark function suites, i.e., CEC-14 and CEC-18, in order to show that selecting the different CEC benchmark suites does not have a crucial impact on estimating the algorithm's quality.