Novel Four-stage Comprehensive Analysis Approach for Population-based Optimization Algorithms

This paper presents a novel four-stage comprehensive analysis approach of population-based optimization algorithms in terms of their performance and versatility of application. The first stage involves the selection of algorithms that do not possess the center-bias operator, i.e., those that do not tend to converge to the center of the search area. In the second and third stages, the accuracy and speed of the selected algorithms are computed, respectively. In the fourth stage, a multi-objective optimization problem is solved, resulting in a list of the most efficient algorithms. The main distinctive feature of this approach is the application at the first stage of the procedure for selecting algorithms that do not possess the center-bias operator, which allows to narrow down the list of algorithms analyzed at the next stages, excluding the obviously defective ones. The proposed approach was used to investigate the capabilities of 10 most popular population-based optimization algorithms in order to select the best of them in terms of performance efficiency and versatility of application. The obtained results can be applied to justify the choice of a particular optimization algorithm for solving a particular applied problem.