Study on MAX–MIN Ant System with Random Selection in Quadratic Assignment Problem

Abstract

Ant Colony Optimization (ACO), which is a type of swarm intelligence inspired by ants' foraging behavior, has been studied extensively and its effectiveness has been shown by many researchers. The previous studies have reported that MAX–MIN Ant System (MMAS) is one of effective ACO algorithms. The MMAS maintains the balance of intensification and diversification concerning pheromone by limiting the quantity of pheromone to the range of minimum and maximum values. In this paper, we propose MAX–MIN Ant System with Random Selection (MMAS^RS) for improving the search performance even further. The MMAS^RS is a new ACO algorithm that is MMAS into which random selection was newly introduced. The random selection is one of the edgechoosing methods by agents (ants). In our experimental evaluation using ten quadratic assignment problems, we have proved that the proposed MMAS^RS with the random selection is superior to the conventional MMAS without the random selection in the viewpoint of the search performance.