Abstract
Local Principal Components Analysis, i.e. Principal Component Analysis performed in data clusters, is discussed and a neural algorithm is developed. This algorithmic tool is used for pattern recognition. A decision function based on the subspace method is generalized by introducing normalization matrix Г and affine coefficients α, β. Assuming that feature measurements are Gaussian in data clusters, it is shown that the new method is equivalent to maximum likelihood method. However, no explicit knowledge on probability distributions of feature vectors in classes, is required. For handwritten numerals the technique reaches the recognition rate of about 99%.
Preview
Unable to display preview. Download preview PDF.
Similar content being viewed by others
References
Hotelling, H.: Analysis of a complex of statistical variables into principal components. Journal of Educational Psychology, 24 1933 417–441
Joutsensalo, J., Miettinen, A.: Self-organizing operator map for nonlinear dimension reduction. ICNN'95, 1995 IEEE International Conference on Neural Networks. 1(1) 1995 111–114
Kambhatla, N., Leen, T.K.: Fast nonlinear dimension reduction. ICNN'93, 1993 International Conference on Neural Networks. 3 1993 1213–1218
Kohonen, T.: The self-organizing map. Proc. of IEEE 78 1990 1464–1480
Kohonen, T.: Self-Organizing Maps. Springer, Berlin, 1995
Linde, Y., Buzo, A., Cray, R.M.: An algorithm for vector quantizer design. IEEE Trans. Comm. COM-28 1980 28–45
Oja E.: Principal components, minor components, and linear neural networks. Neural Networks 5 1992 927–935
Oja E.: Subspace methods of pattern recognition. Research Studies Press, England, 1983
Diamantaras, K.I., Kung, S.Y.: Principal component neural networks. John Wiley & Sons, New York, 1996
Skarbek, W.: Local Principal Components Analysis for Transform Coding. 1996 Int. Symposium on Nonlinear Theory and its Applications, NOLTA'96 Proceedings, Research Society NTA, IEICE, Japan, Oct. 1996 381–384
Suen, C.Y., Legault, R., Nadal, C., Cheriet, M., Lam, L.: Building a new generation of handwriting recognition systems. Pattern Recognition Letters, 14 1993 303–315
Watanabe, S., Pakvasa, N.: Subspace method of pattern recognition. Proc. of 1st Int. Joint Conf. on Pattern Recognition, 1973
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1997 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Skarbek, W., Ghuwar, M., Ignasiak, K. (1997). Local subspace method for pattern recognition. In: Sommer, G., Daniilidis, K., Pauli, J. (eds) Computer Analysis of Images and Patterns. CAIP 1997. Lecture Notes in Computer Science, vol 1296. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-63460-6_159
Download citation
DOI: https://doi.org/10.1007/3-540-63460-6_159
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-63460-7
Online ISBN: 978-3-540-69556-1
eBook Packages: Springer Book Archive