Título: The Effect of the Heuristic Information on a Hybrid MAX−MIN Ant System for the Quadratic Assignment Problem
Autores: Dantas, Augusto;Pozo, Aurora
The MAX-MIN Ant System is a bio-inspired metaheuristic algorithm that has presented some of the best results for combinatorial optimization problems, specially the Quadratic Assignment Problem. But differently from most of the Ant Colony Optimization algorithms, it does not make use of any heuristic information when applied with local search. The rationale for this is that both the heuristic and the local search serve to the same purpose of preventing bad solutions to update the pheromone trails. In this work, we analyse this statement and compare the results obtained with and without heuristic, also determining which pheromone update choice is more suitable for each case. In the experiments, it was observed a clear benefit of using the heuristic information regarding the expended time to find good solutions, with no overall advantage of not using it.
Ant Colony Optimization;Quadratic Assignment Problem;Swarm Intelligence
Código DOI: 10.21528/CBIC2017-44
Artigo em pdf: cbic-paper-44.pdf
Arquivo BibTeX: cbic-paper-44.bib