Abstract
Clusters in data clustering should be robust to sample fluctuation, i.e., the estimate of cluster parameters on a second sample set should yield qualitatively similar results. This robustness requirement can be quantified by large deviation arguments from statistical learning theory. We use the principle of Empirical Risk Approximation to determine an optimal number of clusters for the case of histogram clustering. The analysis validates stochastic approximation algorithms like Markov Chain Monte Carlo which maximize the entropy for fixed optimization costs.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Duda, R.O. and Hart, P.E. (1973): Pattern Classification and Scene Analysis. Wiley, New York.
De Bruijn, N.G.(1958): Asymptotic Methods in Analysis. North-Holland Publishing Co., (repr. Dover), Amsterdam.
Hofmann, T. and Puzicha, J. (1998): Statistical models for co-occurrence data. AI-MEMO 1625, Artifical Intelligence Laboratory. Massachusetts Institute of Technology.
Pereira, F.C.N., Tishby, N.Z. and Lee, L (1993): Distributional clustering of english words. In 30th Annual Meeting of the Association for Computational Linguistics, Columbus, Ohio, pages 183–190.
Puzicha, J., Hofmann, T. and Buhmann, J.M. (1999): Histogram Clustering for Unsupervised Segmentation and Image Retrieval. Pattern Recognition Letters, 20, 135–142.
Van Der Vaart, A.W. and Wellner, J.A. (1996): Weak Convergence and Empirical Processes.Springer-Verlag, New York.
Vapnik, V.N. and Chervonenkis, A.Ya. (1971): On the uniform convergence of relative frequencies of events to their probabilities. Theory of Probability and its Applications, 16, 264–280.
Vapnik, V. N. (1998): Statistical Learning Theory. Wiley-Interscience, New York.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2002 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Buhmann, J.M., Held, M. (2002). On the Optimal Number of Clusters in Histogram Clustering. In: Gaul, W., Ritter, G. (eds) Classification, Automation, and New Media. Studies in Classification, Data Analysis, and Knowledge Organization. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-55991-4_4
Download citation
DOI: https://doi.org/10.1007/978-3-642-55991-4_4
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-43233-3
Online ISBN: 978-3-642-55991-4
eBook Packages: Springer Book Archive