Fast Spectral Clustering for Large Data Sets Using Minimal Enclosing Ball
Hua Xu · Dianzi xuebao · 2010
Graph-based relaxed clustering(GRC) is a spectral clustering algorithm with straightforwardness and adaptability.However,for large data sets,its heavy time cost severely weakens its usefulness.In order to overcome this shortcoming,a novel algorithm named CCMEB-based constrained GRC(CCMEB-CGRC) is proposed in this study by constraining the clustering indicator of GRC and introducing Center-Constrained Minimal Enclosing Ball(CCMEB).This algorithm has the merit of asymptotic linear time complexity as well as inherits the straightforwardness and adaptability of GRC.Thus,a fast and efficient spectral clustering method fitting for large data sets is proposed.This is confirmed in the experimental studies.