Abstract:

The purpose of this article is to study the applicability of neural technologies and fuzzy-multiple algorithms for the management of capital adequacy of a commercial bank. On the basis of integration of financial and mathematical methods of analysis in the article were developed intelligent modules of decision support, which can be built into the banking information system. The author's approbation using the example of one of the Russian commercial banks showed that neural technologies, supplemented by calculation methods based on fuzzy logic, allow to obtain more accurate results which are necessary to support decision-making in the management of capital adequacy calculation, as well as in general management of banking activities. On the basis of the materials set out in the article the following main conclusions are obtained:

• The use of modern neural technologies has certain limitations due to the opacity of the calculation mechanisms and the sufficient complexity of building an artificial neural network. But the financial sector in general and the banking sector in particular have been using regression prediction algorithms to support decision-making for a long time. Therefore, the introduction of neural technologies in the management of capital adequacy of a commercial bank is reasonably expedient;
• It is necessary to integrate neural technologies and fuzzy-multiple calculation algorithms. This allows approximate the quality of the processes in the artificial neural network to the biological analogue. Consequently, fuzzy-multiple algorithms in an artificial neural network are similar to decision making in conditions of uncertainty, limited (incomplete) information, and also in the conditions when some information cannot be presented by concrete quantitative or monetary values;
• Intelligent Decision Support module, which is built on neural technologies and fuzzy-multiple algorithms, can not only assess the state of certain parameters of banking activity (for example, in the area of management Capital adequacy), but also capable of training (as any neural technology). Since the calculations are based on fuzzy logic, the training of the neural model is sufficiently fast and the accuracy of the evaluation results is higher.