Vestnik of Samara State Technical University. Technical Sciences SeriesVestnik of Samara State Technical University. Technical Sciences Series1991-85422712-8938Samara State Technical University2029910.14498/tech.2017.4.%uResearch ArticleTWO-STAGE NORMALIZATION OF OUTPUT SIGNALS OF ARTIFICIAL NEURAL NETWORKSGirinRoman VLead Software Engineer.romangirin@gmail.comOrlovSergey P(Dr. Sci. (Techn.)), Professor.-OOO ‘IntellectSof't’Samara State Technical University1512201725471610022020Copyright © 2017, Samara State Technical University2017In the article the problem of interpretation output signals of artificial neural networks was considered in cases when softmax regression is used in them which calculated with heuristics often applied in practice. To use alternative multi-fold activation function in artificial neural network for objects categorizing is proposed. It is shown that output signals, that can be get in case of using the function, possible to be interpreted as probability of full group of events. Additionally, the form of distribution allows to estimate how precise the proposition is. With aim of developed artificial neural net that recognizes hand-written characters from 0 through 9 experiments were made that checks proposed approach.categorizationartificial neural networksnormalization of valuesкатегоризация объектовискусственные нейронные сетинормализация[Norvig P., Rassell S. Artificial Intelligence: A Modern Approach, Edition: 3rd, Pearson, 2010.][Nielsen M. Neural Networks and Deep Learning, free online book. - http://neuralnetworksanddeeplearning.com, 2017.][Dodge Y. The Oxford Dictionary of Statistical Terms, OUP, 2003.][Goodfellow I., Bengio Y., Courville A. Deep learning. - http://www.deeplearningbook.org, 2017.][Мак-Каллок У.С., Питтс В. Логическое исчисление идей, относящихся к нервной активности // Автоматы / Под ред. К.Э. Шеннона и Дж. Маккарти. - М.: Изд-во иностр. лит., 1956. - С. 363-384.][Grother P., Hanaoka K. NIST Special Database 19 Handprinted Forms and Characters, 2nd Edition, National Institute of Standards and Technology, September 13, 2016.][Rumelhart D.E., Hinton G.E., Williams R.J. Learning Internal Representations by Error Propagation // Parallel Distributed Processing. Vol. 1. - Cambridge, MA: MIT Press, 1986.][Комарцова Л.Г., Максимов А.В. Нейрокомпьютеры. - М.: МГТУ им. Н.Э. Баумана, 2004.][Галушкин А.И. Синтез многослойных систем распознавания образов. - М.: Энергия, 1974.][Барцев С.И., Охонин В.А. Адаптивные сети обработки информации. - Красноярск: Ин-т физики СО АН СССР, 1986.]