A new parameter adaptation method for Genetic Algorithms and Ant Colony Optimization algorithms

This Ant Colony Optimization algorithms and Genetic Algorithms are actively used in controller design, robotic path planning, design automation, biomedical imaging, data mining, and distribution network planning. This paper introduces a genetic algorithm implementation, an ant colony optimization algorithm implementation, and a method of adapting the parameters for the algorithms during the course of their execution whenever they cease producing better solutions. Additionally, it presents the results of experiments performed with and without the method applied. The obtained research outcomes clearly show that the method has the great potential to improve the solutions arrived at in both types of nature inspired algorithms, though the greater improvement is achieved whenever an algorithm tends to stagnate further from the theoretical optimum as happened with the genetic algorithm as compared to with the ant colony optimization algorithm.

I. Introduction

The Traveling Salesman Problem (TSP) is one of the most studied problems in the class of NP Hard problems 1 because it is a good analog for a broad group of real world problems including planning, logistics, microchip manufacturing, and DNA sequencing [2]. We chose to test our hypothesis on the TSP to ease comparison to other research in heuristic optimization. Additionally, the widely used TSPLIB [3] includes many problem instances used in other research, all of which have deterministically determined optimum values available.