A parallel hybrid GA-PSO approach with dynamic rule-based parameter setting
Abtin Nourmohammadzadeh, Sven Hartmann, Hui Ma · Proceedings of the Genetic and Evolutionary Computation Conference Companion · 2017
We present an new optimisation framework combining two meta-heuristics: Genetic Algorithms (GA) and Particle Swarm Optimisation (PSO). In contrast to the usual hybridisation models in which the second algorithm is applied to work on the final results of the first one, our approach uses both algorithms in parallel on the same population in a competetive manner. The algorithms can work on and improve the solutions of each other, thus more diversity and better quality can be achieved in the population. Another improving factor is resetting the population size and the parameters in every iteration according to the diversity and quality of the solutions in the last population. Our approach is tested on five well-known benchmark problems. The merit of our approach is verified by comparing its performance with the pure GA and PSO, hybrids where PSO works after GA, and vice versa, as well as another hybrid approach of these algorithms from the literature.