Partial-ACO as a GA mutation operator applied to TSP instances

Abstract

A recent novel modification to Ant Colony Optimisation (ACO) known as Partial-ACO can be successfully used to solve Travelling Salesman Problems (TSP) by making partial modifications. The approach also dispenses with a pheromone matrix using the population to build pheromone levels on edges enabling scaling to large problems. Consequently, being population based the approach can be also used within a Genetic Algorithm as a mutation operator. Results demonstrate significant improvements when using Partial-ACO as a mutation operator with a range of crossover operators.