Discrete - time hopfield neural network based text clustering algorithm

Abstract

In this study we propose a discrete-time Hopfield Neural Network based clustering algorithm for text clustering for cases L = 2(q) where L is the number of clusters and q is a positive integer. The optimum general solution for even 2-cluster case is not known. The main contribution of this paper is as follows: We show that i) sum of intra-cluster distances which is to be minimized by a text clustering algorithm is equal to the Lyapunov (energy) function of the Hopfield Network whose weight matrix is equal to the Laplacian matrix obtained from the document-by-document distance matrix for 2-cluster case; and ii) the Hopfield Network can be iteratively applied to text clustering for L = 2(k). Results of our experiments on several benchmark text datasets show the effectiveness of the proposed algorithm as compared to the k-means.