An Initialization Method for the Kmeans Algorithm using RNN and Coupling Degree

Abstract

Since K-means is widely used for general clustering, its performance is a critical point. This performance depends highly on initial cluster centers since it may converge to numerous local minima. In this paper a proposed initialization method to select initial cluster centers for K-means clustering is proposed. This algorithm is based on reverse nearest neighbor (RNN) search and coupling degree. Reverse nearest neighbor search retrieves all points in a given data set whose nearest neighbor is a given query point, where coupling degree between neighborhoods of nodes is defined based on the neighborhood-based rough set model as the amount of similarity between objects. The initial cluster centers computed using this methodology are found to be very close to the desired cluster centers for iterative clustering algorithms. The application of the proposed algorithm to K-means clustering algorithm is demonstrated. An experiment is carried out on several popular datasets and the results show the advantages of the proposed method. General Terms Artificial intelligence, Data clustering, partitioning methods, Kmeans Initialization methods