An immune spectral clustering algorithm

Abstract

A new clustering approach namely immune spectral clustering algorithm (ISCA) is proposed in this paper. It combines spectral clustering with immune algorithm for data clustering. In this algorithm, making use of the dimension reduction ability of the spectral clustering algorithm, an immune clonal clustering algorithm is used to cluster the data points in the mapping space. Because we can get tight clusters after mapping with the spectral clustering, and the immune clonal clustering algorithm characterized by rapid convergence to global optimum and minimal sensitivity to initialization, we can get a better data clustering. Experimental results over four data sets from UCI database show the efficiency of our algorithm.