Abstract
Kohonen’s Self-Organizinig Map (SOM) is useful to make a low dimensional manifold which is embedded into a high dimensional data space, hence it is now being used in visualization, analysis, and knowledge discovery of complex data. However, for a given set of training samples, a trained SOM is not uniquely determined because of many local minima, resulting that we can not evaluate whether it is appropriately embedded or not. In this paper, we propose a new method how to evaluate a trained SOM from the viewpoint of geometric naturalness. The new criterion is defined by the average correspondence gap between different trained SOMs. We show the effectiveness of the proposed method by experimental results for artificial data, and then introduce its application to a real-world problem, analysis of cities and villages in Japan.
This is a preview of subscription content, log in via an institution.
Buying options
Tax calculation will be finalised at checkout
Purchases are for personal use only
Learn about institutional subscriptionsPreview
Unable to display preview. Download preview PDF.
References
Adorno, M.C., Resta, M.: Reliability and Convergence on Kohonen Maps: An Empirical Study. In: Negoita, M.G., Howlett, R.J., Jain, L.C. (eds.) KES 2004. LNCS (LNAI), vol. 3213, pp. 426–433. Springer, Heidelberg (2004)
Barron, A.R.: Universal Approximation Bounds for Superpositions of a Sigmoidal Function. IEEE Transactions on Information Theory 39(3) (1993)
de Bodt, E., Cottrell, M., Verleysen, M.: Statistical tools to assess the reliability of self-organizing maps. Neural Networks 15, 967–978 (2002)
Fukui, K.-I., Numao, M.: Topographic Measure Based on External Criteria for Self-Organizing Map. In: Laaksonen, J., Honkela, T. (eds.) WSOM 2011. LNCS, vol. 6731, pp. 131–140. Springer, Heidelberg (2011)
Kohonen, T.: Self-Organizing maps. Springer (1995)
Ohara, S., Watanabe, S.: Correspondence of Different Self-Organizing Maps and its Application to Learning Evaluation. IEICE Technical Report, NC2012-149, 89–94 (2013)
Rumelhart, D.E., McClelland, J.L.: PDP Research Group. Parallel Distributed Processing. MIT Press (1986)
Saitoh, F., Watanabe, S.: On Generalization Error of Self-Organizing Map. In: Wong, K.W., Mendis, B.S.U., Bouzerdoum, A. (eds.) ICONIP 2010, Part II. LNCS, vol. 6444, pp. 399–406. Springer, Heidelberg (2010)
Watanabe, S.: Algebraic Geometry and Statistical Learning Theory. Cambridge University Press (2009)
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2013 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Ohara, S., Yamazaki, K., Watanabe, S. (2013). A Geometric Evaluation of Self-Organizing Map and Application to City Data Analysis. In: Ramanna, S., Lingras, P., Sombattheera, C., Krishna, A. (eds) Multi-disciplinary Trends in Artificial Intelligence. MIWAI 2013. Lecture Notes in Computer Science(), vol 8271. Springer, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-44949-9_16
Download citation
DOI: https://doi.org/10.1007/978-3-642-44949-9_16
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-642-44948-2
Online ISBN: 978-3-642-44949-9
eBook Packages: Computer ScienceComputer Science (R0)