Consultant-Guided Search - A New Metaheuristic for Combinatorial Optimization Problems

摘要

In this paper, we present Consultant-Guided Search (CGS), a new metaheuristic for combinatorial optimization problems, based on the direct exchange of information between individuals in a population. CGS is a swarm intelligence technique inspired by the way real people make decisions based on advice received from consultants. We exemplify the application of this metaheuristic to a specific class of problems by introducing the CGS-TSP algorithm, an instantiation of CGS for the Traveling Salesman Problem. To determine if our direct communication approach can compete with stigmergy-based methods, we compare the performance of CGS-TSP with that of Ant Colony Optimization algorithms. Our experimental results show that the solution quality obtained by CGS-TSP is comparable with or better than that obtained by Ant Colony System and MAX-MIN Ant System.