Indicator Based Ant Colony Optimization for Multi-objective Knapsack Problem

The use of metaheuristics to solve multi-objective optimization problems (MOP) is a very active research topic. Ant Colony Optimization (ACO) has received a growing interest in the last years for such problems. Many algorithms have been proposed in the literature to solve different MOP. This paper presents an indicator-based ant colony optimization algorithm called IBACO for the multi-objective knapsack problem (MOKP). The IBACO algorithm proposes a new idea that uses binary quality indicators to guide the search of artificial ants. These indicators were initially used by Zitzler and Künzli in the selection process of their evolutionary algorithm IBEA. In this paper, we use the indicator optimization principle to reinforce the best solutions by rewarding pheromone trails. We carry out a set of experiments on MOKP benchmark instances by applying the two binary indicators: epsilon indicator and hypervolume indicator. The comparison of the proposed algorithm with IBEA, ACO and other state-of-the-art evolutionary algorithms shows that IBACO is significantly better on most instances.