Maximin Initialization for Cluster Analysis

Abstract

Most iterative clustering algorithms require a good initialization to achieve accurate results. A new initialization procedure for all such algorithms is given that is exact when the data contain compact, separated clusters. Our examples use c-means clustering. 
DOI: 10.1007/11892755_2

Topics

4 Figures and Tables

Cite this paper

@inproceedings{Hathaway2006MaximinIF, title={Maximin Initialization for Cluster Analysis}, author={Richard J. Hathaway and James C. Bezdek and Jacalyn M. Huband}, booktitle={CIARP}, year={2006} }