Article 2118

Title of the article

CORRELATION OF NEURON POWER WITH LINEAR AND SQUARE COMPARISON OF BIOMETRIC DATA 

Authors

Volchikhin Vladimir Ivanovich, Doctor of engineering sciences, professor, the President of Penza State University (40 Krasnaya street, Penza, Russia), 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), ivan@pniei.penza.ru
Malygina Elena Aleksandrovna, Candidate of engineering sciences, research worker, the interindustrial testing laboratory of biometric devices and technologies, Penza State University (40 Krasnaya street, Penza, Russia), mal890@yandex.ru
Yunin Aleksey Petrovich, Lead expert, Penza Research Institute of Electrical Engineering (9 Sovetskaya street, Penza, Russia), alexey_82@mail.ru

Index UDK

519.24; 53; 57.017

DOI

10.21685/2072-3059-2018-1-2

Abstract

Background. The goal of the research is comparative assessment of capacity of conventional neurons with linear summation and neurons with the quadratic sum of standardized and centered biometric data.
Materials and methods. Training of conventional neurons converter biometrics in code is absolutely resistant to clone branch algorithms (state standart GOST R 52633.5-2011). Algorithms for training radial-basic neurons have the same stability.
In terms of stability, there is no difference between training neurons with linear and quadratic enrichment of data.
Results. It is shown that the probabilities of errors of the first and second kind in the processing of biometric data by ordinary neurons are much greater than in the treatment of radial-basic neurons. The slope of the power growth line of quadratic
neurons, depending on the dimension of the problem, has a slope several times steeper than for linear neurons. There is a very rapid increase in the power of quadratic neurons with an increase in the number of entries they have.
Conclusions. High power quadratic neurons makes them extremely promising for application in neural network converters biometrics. Efforts need to be directed to develop special measures to eliminate the disadvantage of quadratic neurons store open mathematical expectation and standard deviation of personal biometric parameters.

Key words

neuronet converter «biometry-code», biometric data, large dimension of data

Download PDF
References

1. Dodis Y., Reyzin L., Smith A. Proc. EUROCRYPT. 2004, April 13, pp. 523–540.
2. Monrose F., Reiter M., Li Q., Wetzel S. Proc. IEEE Symp. on Security and Privacy. 2001, pp. 202–213.
3. Hao F., Anderson R., Daugman J. IEEE transactions on computers. 2006, vol. 55, no. 9, pp. 1073–1074.
4. Yazov Yu. K., Volchikhin V. I., Ivanov A. I., Funtikov V. A., Nazarov I. G. Neyrosetevaya zashchita personal'nykh biometricheskikh dannykh [Neural network protection of personal biometric data]. Moscow: Radiotekhnika, 2012, 157 p.
5. Ivanov A. I. Biometricheskaya identifikatsiya lichnosti po dinamike podsoznatel'nykh dvizheniy: monografiya [Biometric identification of a person by the dynamics of subconscious movements: monograph]. Penza: Izd-vo PGU, 2000, 178 p.
6. Gudfellou Ya., Bendzhio I., Kurvil' A. Glubokoe obuchenie [In-depth traning]. Moscow: DMK Press, 2017, 652 p.
7. GOST R 52633.5–2011. Zashchita informatsii. Tekhnika zashchity informatsii. Avtomaticheskoe obuchenie neyrosetevykh preobrazovateley biometriya-kod dostupa [State Standard GOST R 52633.5-2011. Information protection. Information protection technique. Automatic learning of neural network "biometrics-access code" converters]. Moscow, 2011.
8. Ivanov A. I. Neyrosetevye tekhnologii biometricheskoy autentifikatsii pol'zovateley otkrytykh sistem: avtoref. dis. d-ra tekhn. nauk: 05.13.01 «Sistemnyy analiz, upravlenie i obrabotka informatsii» [Neural network technologies for biometric authentication of
open systems users: author's abstract of dissertation to apply for the degree of the candidate of medical sciences: 05.13.01 «System analysis, management and information processing»]. Penza, 2002, 34 p.
9. Volchikhin V. I., Ivanov A. I., Funtikov V. A., Malygina E. A. Izvestiya vysshikh uchebnykh zavedeniy. Povolzhskiy region. Tekhnicheskie nauki [University proceedings. Volga region. Engineering sciences]. 2013, no. 4 (28), pp. 86–96.
10. Marshalko G. B. Matematicheskie voprosy kriptografii [Mathematical problems of cryptography]. 2014, vol. 5, no. 2, pp. 87–98.
11. 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.
12. Ivanov A. I., Perfilov K. A. Sovremennye okhrannye tekhnologii i sredstva obespecheniya kompleksnoy bezopasnosti ob"ektov: sb. materialov XI Vseros. NPK [Modern security technologies and facilities for providing complex security of facilities: proceedings
of XI All-Russian scientific and practical conference]. Penza; Zarechnyy, 2016, pp. 223–229.
13. Ivanov A. I., Perfilov K. A., Malygina E. A. Vestnik Sibirskogo gosudarstvennogo aerokosmicheskogo universiteta im. akademika M. F. Reshetneva [Bulletin of Siberian State Aerospace University]. 2016, no. 4 (17), pp. 864–871.

 

Дата создания: 13.06.2018 14:02
Дата обновления: 03.07.2018 16:15