Different approaches in parallelization of the artificial bee colony algorithm
Miloš Subotić, Milan Tuba, Nadezda Stanarevic · 2011
Parallel processing is gaining popularity due to the low cost of multi-core processors. In this paper we propose three different approaches in parallelization of standard artificial bee colony (ABC) algorithm. ABC algorithm was successfully used on many optimization problems, unconstrained and constrained. Our three approaches are independent parallel runs and two variations of multiple swarms parallelization. By using independent parallel runs method we succeeded in achieving faster execution of algorithm since multicore processors can be better utilized. By using multiple swarms technics with some modifications we also obtained better results than the original ABC algorithm. Different types of communications among swarms are proposed and examined. These methods of communication between swarms improved results and allowed adjustments of different ratios between exploration and exploitation. Set of eleven standard benchmark functions was used to test execution speed and quality of results improvements. Keywords—Artificial bee colony, Metaheuristic optimization, Parallelization, Swarm intelligence, Nature inspired metaheuristic algorithms.