UDC 004.32.26
ON APPLICATION OF DEEP NEURON NETWORKS TO EVALUATE ENTERPRISE CREDIT CAPACITY
K. G. Shitova, student RSREU; Ryazan, Russia; This email address is being protected from spambots. You need JavaScript enabled to view it.
N. I. Tsukanova, Ph.D. (Tech.), Associate professor, department of computational and applied mathematics, RSREU, Ryazan, Russia; This email address is being protected from spambots. You need JavaScript enabled to view it.
The aim of the article is to study the application of neuron networks while evaluating small business enterprise credit capacity. The issues of multilevel neuron network learning are considered. Different methods to increase the accuracy, performance and stability of the models received are described and discussed. The attention is also given to the development of applications both in MatLab environment and C# language that implement different algorithms of multilevel neuron networks learning. Consequently, the work considers the following aspects: the task to evaluate small business credit capacity and the possibility to solve it by means of machine learning, the task of multilevel neuron network learning as the task of definite dynamic process optimization; the problems arising in the process of multilevel neuron network learning as well as the ways to solve them. The results of the research made with the help of the applications developed are given, the recommendations for their usage while evaluating small business credit capacity are shown.
Key words: credit capacity evaluation, multilevel neuron network, deep neuron networks, error propagation algorithm, step and batch learning modes, aim function, activation functions, learning error, generalization or testing error.