Enhancing The Behavior Of The Ant Algorithms In Solving Optimization Problems

Abstract

In recent years, Swarm Intelligence , and in particular, Ant algorithms have received much attention among researchers as promising search and optimization techniques which are inspired from nature. Ant algorithms depend essentially on the idea of synergistic use of cooperation among many relatively simple agents, which communicate by distributed memory. In such a research area, several approaches have been proposed to simulate the behavior of real ants. Among such approaches, Ant Colony System (ACS) comes as the most interesting one.

The aim of this thesis is to introduce two enhancements to Ant Colony System, which is considered one of the most successful ant algorithms used to solve combinatorial optimization problems. The thesis presents the results of 15 experiments, where the proposed algorithm is compared with the classical Ant Colony System in solving shortest path problems. Experimental work show that the modified ACS algorithm outperforms the classical one in terms of reducing the number of tours needed to reach the optimum solution and increasing the ability of dealing with different instances of the shortest path problem.