Reference-Point Based Multi-swarm Algorithm for Many-Objective Problems

Many-Objective Optimization Problems (MaOPs) are problems that have more than three objectives to be optimized. Usually, the state-of-art of Multi-Objective Evolutionary algorithms scale poorly when the number of objective functions increases. To overcome this limitation, researches are investigating multi-swarm approaches. Besides, another newly strategy is the use of reference points to enhance the search of the algorithms. Based on those strategies, this work proposes a new multi-swarm algorithm, called Reference-Point Based Multi-Swarm Algorithm, R-Multi, which takes advantages of reference points to guide a multi-swarm search. The main idea is to use reference points to guide the search towards the Pareto front and to perform the communication between swarms allowing the necessary collaboration to have an effective exploration of the search space. Furthermore, this work presents a set of experiments that compare R-Multi to others multi-swarm algorithms and to MOEA/D-DRA. The algorithms are evaluated in several MaOPs observing both convergence and diversity. The results shows the validity of the proposed algorithm and stresses the good results of multi-swarm approaches in Many-Objective Optimization.