A new variant of the Particle Swarm Optimization (PSO) algorithm is presented in this paper. It uses a well-known measure of problem hardness, the Fitness-Distance Correlation, to modify the position of the swarm attractors, both global and local to single particles. The goal of the algorithm is to make the fitness landscape between each particle's positions and their attractors as smooth as possible. Experimental results, obtained on 15 out of the 25 test functions belonging to the test suite used in CEC-2005 numerical optimization competition, show that this new PSO version is generally competitive, and in some cases, better than standard PSO.
FDC-based particle swarm optimization / A. Azzini, S. Cagnoni, L. Vanneschi - In: Artificial life and evolutionary computation : proceedings of Wivace 2008 : Venice, Italy, 8–10 september 2008 / [a cura di] R. Serra, M. Villani, I. Poli. - Singapore : World scientific, 2009. - ISBN 9789814287449. (( Intervento presentato al 2. convegno Workshop Italiano di Vita Artificiale e Computazione Evolutiva (WIVACE) tenutosi a Venezia nel 2008.
FDC-based particle swarm optimization
A. AzziniPrimo
;
2009
Abstract
A new variant of the Particle Swarm Optimization (PSO) algorithm is presented in this paper. It uses a well-known measure of problem hardness, the Fitness-Distance Correlation, to modify the position of the swarm attractors, both global and local to single particles. The goal of the algorithm is to make the fitness landscape between each particle's positions and their attractors as smooth as possible. Experimental results, obtained on 15 out of the 25 test functions belonging to the test suite used in CEC-2005 numerical optimization competition, show that this new PSO version is generally competitive, and in some cases, better than standard PSO.Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.