iMOACOR: A New Indicator-Based Multi-Objective Ant Colony Optimization Algorithm for Continuous Search Spaces


Abstract

Ant colony optimization (ACO) is a metaheurisitc which was originally designed to solve combinatorial optimization problems. In recent years, ACO has been extended to tackle continuous single-objective optimization problems, being ACOR one of the most remarkable approaches of this sort. However, there exist just a few ACO-based algorithms designed to solve continuous multi-objective optimization problems (MOPs) and none of them has been tested with many-objective problems (i.e., multi-objective problems having four or more objectives). In this paper, we propose a novel multi-objective ant colony optimizer (called iMOACOR) for continuous search spaces, which is based on ACOR and the R2 performance indicator. Our proposed approach is the first specifically designed to tackle many-objective optimization problems. Moreover, we present a comparative study of our proposal with respect to NSGA-III, MOEA/D, MOACOR and SMS-EMOA using standard test problems and performance indicators adopted in the specialized literature. Our preliminary results indicate that iMOACOR is very competitive with respect to state-of-the-art multi-objective evolutionary algorithms and is also able to outperform MOACOR.