Article 5318

Title of the article

COMPARISON OF CAPACITIES OF TWO TYPES OF ARTIFICIAL NEURONS ENRICHING BIOMETRIC
DATA IN LINEAR AND QUADRATIC SPACES 

Authors

Volchikhin Vladimir Ivanovich, Doctor of engineering sciences, professor, president of Penza State University (40 Krasnaya street, Penza, Russia), E-mail: president@pnzgu.ru
Ivanov Aleksandr Ivanovich, Doctor of engineering sciences, associate professor, head of the laboratory of biometric and neural network technologies, Penza Research Institute of Electrical Engineering (9 Sovetskaya street, Penza, Russia), E-mail: ivan@pniei.penza.ru
Malygina Elena Aleksandrovna, Candidate of engineering sciences, researcher, interdisciplinary laboratory of biometric devices and technologies testing, Penza State University (40 Krasnaya street, Penza, Russia), E-mail: mal890@yandex.ru
Serikova Yuliya Igorevna, Postgraduate student, Penza State University (40 Krasnaya street, Penza, Russia), E-mail: julia-ska@yandex.ru 

Index UDK

519.24; 53; 57.017 

DOI

10.21685/2072-3059-2018-3-5 

Abstract

Background. The aim of the work is to numerically compare the potential of two types of artificial neurons engaged in enriching input biometric data in linear and quadratic space.
Materials and methods. Numerical simulation is used for two types of neurons, built on simple geometric models of monotonously increasing dimension, operable with symmetrization of correlation links of high-dimensional biometric data.
Results. It is logically proved that networks of artificial neurons comparable in capacity with a summation in linear space should have more than twice the number of neurons compared to networks of neurons that enrich data in quadratic spaces. This effect increases with the growth of dimension of the problem solved by neural networks. It is weakly affected by the natural correlation of real biometric data.
Conclusions. The capacity of neural networks trained in accordance with GOST R 52633.5 can be enhanced if we move from ordinary neurons to hybrid neurons that enrich the data simultaneously in both linear and quadratic spaces. At the same time, the computational complexity of learning algorithms for hybrid neurons should increase from linear to polynomial (quadratic). 

Key words

correlation coefficients, artificial neurons (perceptrons), artificial neurons of quadratic forms, high dimensionality of statistical analysis 

 Download PDF
References

1. GOST R 52633.0–2006. Zashchita informatsii. Tekhnika zashchity informatsii. Trebovaniya k sredstvam vysokonadezhnoy biometricheskoy autentifikatsii [Information protection. Information protection technology. Requirements to the means of highreliability biometric authentication]. Moscow, 2006.
2. GOST R 52633.5–2011. Zashchita informatsii. Tekhnika zashchity informatsii. Avtomaticheskoe obuchenie neyrosetevykh preobrazovateley biometriya-kod dostupa [Information protection. Information protection technology. Automatic learning of neural network biometrics-access code converters]. Moscow, 2011.
3. Tekhnicheskaya spetsifikatsiya (proekt, publichnoe obsuzhdenie vtoroy redaktsii s 01.06.2018 v TK 26 «Kriptograficheskaya zashchita informatsii»). Zashchita neyrosetevykh biometricheskikh konteynerov s ispol'zovaniem kriptograficheskikh algoritmov [Technical specification (draft, public discussion of the second edition on 01.06.2018 in TK 26 “Cryptographic protection of information”). Protection of neural network biometric containers using cryptographic algorithms]. Moscow, 2018.
4. O personal'nykh dannykh: feder. zakon № 152-FZ ot 27 iyunya 2006 g. [On personal data: the federal law № 152-FZ of June 27th, 2006]. Available at: https://duma.consultant.ru/page.aspx?878610
5. Volchikhin V. I., Ivanov A. I., Vyatchanin S. E. Novye informatsionnye tekhnologii i sistemy: sb. nauch. st. XIV Mezhdunar. nauch.-tekhn. konf. (Penza 22–24 noyabrya, 2017) [New information technologies and system: proceedings of XIV International scientific and technical conference (Penza, November 22nd-24th, 2017)]. Penza, 2017, pp. 159–163.
6. Volchikhin V. I., Ivanov A. I., Vyatchanin S. E., Malygina E. A. Izvestiya vysshikh uchebnykh zavedeniy. Povolzhskiy region. Tekhnicheskie nauki [University proceedings. Volga region. Engineering sciences]. 2017, no. 2 (42), pp. 55–65.
7. GOST R 52633.3–2011. Zashchita informatsii. Tekhnika zashchity informatsii. Testirovanie stoykosti sredstv vysokonadezhnoy biometricheskoy zashchity k atakam podbora [Information protection. Information protection technology. Testing the resistance of high reliability biometric protection means to picking attacks]. Moscow, 2011.
8. GOST R 52633.1–20009. Zashchita informatsii. Tekhnika zashchity informatsii. Trebovaniya k formirovaniyu baz estestvennykh biometricheskikh obrazov, prednaznachennykh dlya testirovaniya sredstv vysokonadezhnoy biometricheskoy autentifikatsii [Information protection. Information protection technology. Requirements to the formation of bases of natural biometric images intended for testing of high reliability biometric authentication means]. Moscow, 2011.
9. GOST R 52633.2–2010. Zashchita informatsii. Tekhnika zashchity informatsii. Trebovaniya k formirovaniyu sinteticheskikh biometricheskikh obrazov, prednaznachennykh dlya testirovaniya sredstv vysokonadezhnoy biometricheskoy autentifikatsii [Information protection. Information protection technology. Requirements to the formation of synthetic biometric images intended for testing og high reliability biometric authentication means]. Moscow, 2011.
10. Ivanov A. I., Zakharov O. S. Sreda modelirovaniya «BioNeyroAvtograf». Programmnyy produkt sozdan laboratoriey biometricheskikh i neyrosetevykh tekhnologiy [Simulation environment “BioNeuroAutograph”. Software created by the laboratory of
biometric and neural network technologies]. Available at: http://pniei.rf/activity/science/noc/bioneuroau-tugraph.zip.
11. Ivanov A. I. Avtomaticheskoe obuchenie bol'shikh iskusstvennykh neyronnykh setey: uchebnoe posobie k paketu laboratornykh rabot «BioNeyroAvtograf» [Automatic learning of large artificial neural networks: tutorial to the practical training package “Bio-NeuroAutograph”]. Penza, 2013, 30 p. Available at: http://pniei.rf/activity/science/noc/tm_ IvanovAI.pdf
12. Akhmetov B. B., Ivanov A. I. Mnogomernye statistiki sushchestvenno zavisimykh biometricheskikh dannykh, porozhdaemye neyrosetevymi emulyatorami kvadratichnykh form: monografiya [Multidimensional statistics of significantly dependent biometric data
created by neural network emulators of quandratic forms: monograph]. Kazakhstan-Almaty: Iz-vo LEM, 2016, 86 p.
13. Volchikhin V. I., Akhmetov B. B., Ivanov A. I. Izvestiya vysshikh uchebnykh zavedeniy. Povolzhskiy region. Tekhnicheskie nauki [University proceedings. Volga region.Engineering sciences]. 2016, no. 1 (37), pp. 5–15.
14. Volchikhin V. I., Ivanov A. I., Malygina E. A., Serikova Yu. I. Izmerenie. Monitoring. Upravlenie. Kontrol' [Measurement. Monitoring. Management. Control]. 2018, no. 1 (23), pp. 66–74.

 

Дата создания: 19.04.2019 14:02
Дата обновления: 22.04.2019 08:13