This paper presents a combination of two wellknown metaheuristic algorithms, Particle Swarm Optimization (PSO) and Ant Colony System (ACS), based on a framework design named A-B-Domain. We take the T ravelling Salesman's Problem as the benckmark problem. ACPS2, as we name this combination, works as a metaheuristic for the TSP. When considering deviations to lower bounds, ACPS2 shows an improvement over the simple ACS with a high computational cost. Proposed policies are able to reduce, significatively, running times. As a final conclusion we observe that a guided search through ACS possible sets of parameters obtains better results than the basic ACS with an extended number of trials. ©2007 IEEE.
|Original language||English (US)|
|Title of host publication||ICIIS 2007 - 2nd International Conference on Industrial and Information Systems 2007, Conference Proceedings|
|Number of pages||6|
|State||Published - Dec 1 2007|