Abstract

ACO prototype system optimization -based k means clustering algorithm research

Author(s): Xin Wang, Zhi Xu,Wei Yuan

Cluster analysis is an importantmethod in image identification, information retrieval, data mining and spatial database research, from which K means algorithmis a kind of clustering algorithmbased on classificationmethod, the algorithm thought is providing K pieces of classification on N pieces of objects, and every classification of them represents a cluster, by comparing every cluster calculated mean and all patterns samples mean, it gets a most similar cluster, constantly repeat such process till objects in cluster all are similar and different clustersÂ’ objects are different, while objective function convergence lets square error function value to be the minimum one.ACO(Ant Colony Optimization)is a kind of simulating ant colony foraging behaviorsÂ’ bio-inspired optimization calculation, due to the algorithm reflects prominent applicability in complex optimization problemsÂ’ solution aspect, let it to get well applied in robot system, picture processing, manufacturing system, vehicle route system and communication system. Therefore, the paper analyzes K means clustering algorithm, it gets the algorithm shortcomings, and uses ACO prototype systemto optimize K means clustering algorithm, and states the algorithm feasibility and superiority.