Estimating an Optimal Neighborhood Size in the Spherical Self-Organizing Feature Map

Authors: Alexandros Leontitsis, Archana P. Sangole

Abstract:

This article presents a short discussion on optimum neighborhood size selection in a spherical selforganizing feature map (SOFM). A majority of the literature on the SOFMs have addressed the issue of selecting optimal learning parameters in the case of Cartesian topology SOFMs. However, the use of a Spherical SOFM suggested that the learning aspects of Cartesian topology SOFM are not directly translated. This article presents an approach on how to estimate the neighborhood size of a spherical SOFM based on the data. It adopts the L-curve criterion, previously suggested for choosing the regularization parameter on problems of linear equations where their right-hand-side is contaminated with noise. Simulation results are presented on two artificial 4D data sets of the coupled Hénon-Ikeda map.

Keywords: Parameter estimation, self-organizing feature maps, spherical topology.

Digital Object Identifier (DOI): doi.org/10.5281/zenodo.1056625

References:

[1] T. Kohonen, "Self-organized formation of topologically correct feature maps", Biological Cybernetics, vol. 43, pp. 59-69, 1982
[2] G. Deboeck, T. Kohonen, Visual Explorations in Finance with Selforganizing Maps. London: Springer-Verlag, 1998
[3] M. Gross, F. Seibert, "Visualization of multi-dimensional image data sets using a neural network", The Visual Computer, International Journal of Computer Graphics, vol. 10, pp. 145-159, 1993
[4] M. Gross, Visual Computing: The Integration of Computer Graphics, Visual Perception and Imaging. Berlin: Springer-Verlag, 1994
[5] J. Vesanto, "SOM-based data visualization methods", Journal of Intelligent Data Analysis, vol. 3, pp. 111-126, 1999
[6] J. Vesanto, E. Alhoniemi, "Clustering of the self-organizing map", IEEE Transactions on Neural Networks, vol. 11, pp. 586-600, 2000
[7] H. Ritter, "Self-organizing maps on non-Euclidean spaces", in Kohonen Maps, E. Oja and S. Kaski, Amsterdam: Elsevier Science B. V., 1999, pp. 97-109
[8] A. Sangole, G. K. Knopf, "Representing high-dimensional data sets as close surfaces", Journal of Information Visualization, vol. 1, pp. 111- 119, 2002
[9] A. P. Sangole, "Data-driven Modeling using Spherical Self-organizing Feature Maps", Ph.D. dissertation, Dept. of Mech. and Mat. Eng., The University of Western Ontario, London, ON, Canada, 2003
[10] A. Sangole, G. K. Knopf, "Geometric representations for highdimensional data using a spherical SOFM", International Journal of Smart Engineering System Design, vol. 5, pp. 11-20, 2003
[11] A. Sangole, G. K. Knopf, "Visualization of random ordered numeric data sets using self-organized feature maps", Computers and Graphics, vol. 27, pp. 963-976, 2003
[12] K. Haese, "Self-organizing feature maps with self-adjusting learning parameters", IEEE Transactions on Neural Networks, vol. 9, pp. 1270- 1278, 1998
[13] P. C. Hansen, Rank-Deficient and Discrete Ill-Posed Problems. SIAM, 1997, pp. 186-193
[14] M. Hénon, "A two-dimensional map with strange attractor", Communications in Mathematical Physics, vol. 50, pp. 69-77, 1976
[15] K. Ikeda, "Multiple-valued stationary state and its instability of the transmitted light by a ring cavity system", Optics Communications, vol. 30, p. 257, 1979