2021, Vol. 6, No. 4. - go to content...

Permanent address of this page - https://kostumologiya.ru/en/15tlkl421.html

Метаданные этой статьи так же доступны на русском языке

Full article in PDF format (file size: 692.6 KB)

For citation:

Smirnov E.E., Kostyleva V.V., Razin I.B., Mironov V.P. Artificial neural networks in solving problems of shoe classification. Journal of Clothing Science. 2021; 6(4). Available at: https://kostumologiya.ru/PDF/15TLKL421.pdf (in Russian).

Artificial neural networks in solving problems of shoe classification

Smirnov Evgeny Evgenievich
Kosygin State University of Russia, Moscow, Russia
E-mail: evg7162@mail.ru
RSCI: https://elibrary.ru/author_profile.asp?id=1097456

Kostyleva Valentina Vladimirovna
Kosygin State University of Russia, Moscow, Russia
E-mail: kostyleva-vv@rguk.ru
RSCI: https://elibrary.ru/author_profile.asp?id=353612

Razin Igor Borisovich
Kosygin State University of Russia, Moscow, Russia
E-mail: razin-ib@rguk.ru
RSCI: https://elibrary.ru/author_profile.asp?id=850439

Mironov Vladislav Petrovich
Kosygin State University of Russia, Moscow, Russia
E-mail: mironov-vp@rguk.ru

Abstract. Nowadays there is a phenomenon called «Information explosion». This term was used to designate the explosive growth of information produced by humanity. Each year the increase is more than 30 % from the previous one. At the same time, all structured and well-formalized data make up no more than 5 %. Based on the rate of growth of information, it becomes clear that a person is not able to structure and formalize a huge portion of data using classical methods. To some extent, algorithms are saving, but they also need processed information. The solution to this problem is neural networks that allow partially imitating the work of the human brain in the field of decision-making. Unlike the algorithmic approach, which gives a specific answer, neural networks work with probabilities, due to which it is possible to work with a classification based on weakly formalized data. There are a large number of classes and types of neural networks, each class is more suitable for solving a certain type of problem. As a basic understanding of the principle of functioning and programming of a neural network, in this article we will consider a fully connected neural network. The principle of its creation, training, testing and operation. This article describes the principle of training an artificial neural network to solve classification problems using the backpropagation method. The purpose of the study is the ability to train a neural network to recognize and classify objects belonging to shoe groups of different age and gender groups and purposes, based on a set of input parameters and/or images. Since the topic is very broad, only the concept of a neural network is demonstrated here. Based on three input parameters, the neural network assigns objects to one of two classes: «Shoes», «Other object». This approach is being developed within the framework of the dissertation for the degree of Doctor of Technical Sciences by Evgeny Evgenyevich Smirnov and is supposed to be introduced into the educational process of the departments «Artistic Modeling, Design and Technology of Leather Products», «Information Technologies» of the Russian State University A.N. Kosygin (Technology. Design. Art) in the form of teaching aids.

Keywords: neural networks; machine learning; artificial intelligence; classification; shoes

Download article in PDF format