Optimizing the Ant Colony Optimization Algorithm Using Neural Network for the Traveling Salesman Problem

Ant Colony Optimization (ACO) has been proved to be one of the most effective algorithms to solve a wide range of combinatorial optimization (or NP-hard) problems as the Travelling Salesman Problem (TSP). The first step of an ACO algorithm is setting the parameters that drive the algorithm. The basic parameters that are used in ACS algorithms are; number of ants, the relative importance (or weight) of pheromone, the relative importance of heuristics value, initial pheromone value, evaporation rate, and a parameter to control exploration or exploitation. Generally these parameters are set as discrete values, but past studies has shown that the behavior of the algorithm can be influenced by the modification of the parameters. In this paper, we propose a Neural Network (NN) to adapt two parameters of the ACO algorithm dynamically. The proposed hybrid algorithm is applied on the classical TSP, and tested on different TSP benchmark instances.