Author(s)

J.S. Pan, S.C. Chu & Z.M. Lu

Abstract

The maximum descent (MD) algorithms have been proposed for clustering objects. Compared with the traditional K-means (or GLA, or LEG) algorithm, the MD algorithms achieve better clustering performance with far less computation time. However, the searching of the optimal partitioning hyperplane of a multidimensional cluster is a difficult problem in the MD algorithms. In this paper, a new partition technique based on tabu search (TS) approach is presented for the MD algorithms. Experimental results show that the tabu search based MD algorithm can produce a better clustering performance than the K-means and MD algorithms. 1 Introduction Clustering plays an important role in data mining, pattern recognition and data compression. One of the main applications

Keywords