Distributed spectral clustering on the coordinator model

Abstract

Clustering is a popular subject in non-supervised learning. Spectral clustering is a method for clustering that reduces dimensionality of data and guarantees a faster convergence to almost optimal clusters. Given a set of n points in Rd , let G = (V, E) be an undirected graph where each vertex represents a point in Rd and there is a nonnegative weighted edge between each pair of vertices (u, v) representing the similarity between u and v.[...]