The use of k-means++ for approximate spectral clustering of large datasets | IEEE Conference Publication | IEEE Xplore