Pattern recognition with the help of quadratic discriminant functions
- 24 Downloads
The problem of pattern recognition with the help of spherical and elliptic discriminant functions is studied; in so doing the pattern of an object is assumed to be a vector of its characters from a finite-dimensional Euclidean space. Using a conformal mapping of a punctured sphere onto the plane as well as the inversion transformation, a criterion for the error-free recognition of two sets containing a finite number of points of training samples is obtained with the help of spherical discriminant functions. An algorithm for solving approximately a problem of construction of an ellipsoid of the minimal volume containing a given finite set of points is described. Bibliography: 5 titles.
KeywordsPattern Recognition Euclidean Space Finite Number Training Sample Discriminant Function
Unable to display preview. Download preview PDF.
- 1.A. L. Kapatskjyan, E. M. Karalova, and Yu. A. Malakyan, “The coefficient of relative position of particles and its use in the quantitative analysis of the distribution of a substance in cells,”Tsytologiya,30, No. 3, 361–365 (1988).Google Scholar
- 2.Yu. V. Boiko, P. G. Litvinenko, Yu. I. Petunin, and K. P. Ganina, “The index of clustering of the chromatin structure in interphase nuclei,”Dop. Akad. Nauk USSR, Ser. A, No. 5, 75–78 (1990).Google Scholar
- 3.K. P. Ganina, Yu. I. Petunin, and Ya. G. Timoshenko, “Quantitative characteristic of nuclear polymorphism of epithelial cells in the case of fibroadenomatosis and cancer of mamma,”Vopr. onkol.,30, No. 12, 1440–1449 (1989).Google Scholar
- 4.G. E. Shilov,Introduction to the theory of linear spaces [in Russian], Gostekhteorizdat, Moscow (1956).Google Scholar
- 5.W. Koppenfels und F. Stallmann,Praxis der Konformen Abbildung [in German], Springer-Verlag, Berlin (1959).Google Scholar