350 rub
Journal Neurocomputers №4 for 2018 г.
Article in number:
The formalization of a mathematical model of a gate with a trained neural network hidden layer
Type of article: scientific article
UDC: 004.056.55:004.272.25
Keywords: gate neural network learning logical basis Boolean algebra n-dimensional hypercube hardware neural network PLD.
Authors:

T.E. Mikhailyuk – Post-graduate Student, Department of Electronics and Biomedical Technology,  Ufa State Aviation Technical University 

E-mail: realotoim@mail.ru

S.V. Zhernakov – Dr.Sc. (Eng.), Professor, Head of Department of Electronics and Biomedical Technology,  Ufa State Aviation Technical University 

E-mail: zhsviit@mail.ru

Abstract:

The influence of the technical environment on the neuron model is considered. The model of the learning Boolean basis is given. A model of a gate neural network with a trained hidden layer is being developed. The method of its integration into the hardware neural network architecture is described. The minimum number of elements of the hidden layer required for an arbitrary mapping in an n-dimensional Boolean space is obtained.

Pages: 18-23
References
  1. Mihajlyuk T.E., ZHernakov S.V. Model' ventil'noj nejronnoj seti i algoritm ee obucheniya // Nejrokomp'yutery: razrabotka, primenenie. 2017.  № 3. S. 27–33.
  2. Mihajlyuk T.E., ZHernakov S.V. Povyshenie ehffektivnosti ispol'zovaniya resursov mikroskhemy PPVM pri realizacii nejronnyh setej // Nejrokomp'yutery: razrabotka, primenenie. 2016. № 11. S. 30–39.
  3. YAblonskij S.V. Vvedenie v diskretnuyu matematiku: Ucheb. posobie dlya vuzov. Izd. 2-e, pererab. i dop. M.: Nauka. 384 s.
  4. Galushkin A.I. Nejromatematika. Kn. 6: Ucheb. posobie dlya vuzov. M.: IPRZHR. 2002. 448 s.
Date of receipt: 27 марта 2018 г.