Stability-based model order selection for clustering using multiple cooperative particle swarms

Abstract

Data clustering is the organization of a set of unlabelled data into similar groups. In this chapter, stability analysis is proposed to determine the model order of the underlying data using multiple cooperative swarms clustering. The mathematical explanations demonstrating why multiple cooperative swarms clustering leads to more stable and robust results than those of single swarm clustering are also provided. The proposed approach is evaluated using different data sets and its performance is compared with that of other clustering techniques. © 2009 Springer-Verlag Berlin Heidelberg.