Kohonen artificial neural networks in the service of a commercial bank

The article considers the substantiation of the alternative application of an artificial neural network as a statistical model to determine the solvency of commercial bank customers. Using the extended Kohonen neural network it is shown how a neural network trained on statistics of previous credit transactions solves the problem of classification of borrowers on the basis of their ability to pay.

References
[1] Galushkina A.I., Tsypkina Ya.I. Neyronnye seti: istoriya razvitiya teorii [Neural Networks: The History of Theory Development]. Moscow, Alyans Publ., 2015, 840 p.
[2] Sokolyanskiy V.V., Zakharov A.V., Nevretdinov A.R., Liev Sh.M. Voprosy ekonomicheskikh nauk - Problems of Economic Sciences, 2015, no. 1, pp. 81-87.
[3] Lavrushin O.I. Bankovskoe delo: sovremennaya sistema kreditovaniya [Banking: the Modern Crediting System]. Moscow, KNORUS Publ., 2013, 360 p.
[4] Hawley D.D., Johnson J.D., Raina D. Financial Analysts Journal, November/December 1990, рр. 63-72.
[5] Altman E.I., Marko G., Varetto F. Journal of Banking and Finance, May 1994, pр. 505-529.
[6] Poddig T. Bankruptcy Prediction: A Comparison with Discriminant Analysis. In Neural Networks in Capital Markets, Refenes A.P., (ed.), New York, John Wiley & Sons Publ., 1994, pр. 311-323.
[7] Kohonen T. Self-Organizing Maps. Heidelberg, New York, 2001, 501 p. [In Russ.: Kohonen T. Samoorganizuyushchiesya karty. Moscow, BINOM, Laboratoriya znaniy Publ., 2009, 624 p.].