Investigating the Influence of Multiple Parameter Settings on Selected Metaheuristic Algorithms

Several optimization algorithms are being proposed by researchers to provide near optimal solution to problems. Some of these algorithms are popular and require much parameter tuning while others are new and require few parameters tuning hence less stressful. There is need for a fair comparison of such optimization algorithms. This study presents the experiment procedure and results on the performance evaluation of four popular metaheuristic algorithms; Genetic Algorithm, Particle Swarm Optimization, Bat Algorithm, Firefly Algorithm, and a new metaheuristic algorithm with less parameter tuning; Symbiotic Organism Search Optimization. The metaheuristic algorithms were evaluated using twelve benchmark functions consisting of six unimodal and six multimodal functions. Experiment results showed that the Symbiotic Organism Search optimization outperformed all the compared algorithms. It is established in the study that the Symbiotic Organism Search optimization with a smaller number of parameters to tune has better performance compared to those requiring high number of parameters to tune.