Article 9116

Title of the article

COMPENSATION OF METHODOLOGICAL ERRORS IN CALCULATIONS OF STANDARD DEVIATIONS AND CORRELATION COEFFICIENTS OCCURING DUE TO SMALL SAMPLE SIZES 

Authors

Volchikhin Vladimir Ivanovich, Doctor of engineering sciences, professor, President of Penza State University (40 Krasnaya street, Penza, Russia), cnit@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
Serikova Yuliya Igorevna, Postgraduate student, Penza State University (40 Krasnaya street, Penza, Russia), julia-ska@yandex.ru

Index UDK

519.7; 519.66; 57.087.1, 612.087.1

Abstract

Background. Enhancement of training algorithms for nearla networks of “biometrics-access code” converters is hindered by methodological errors occurring due to a small number of examples in training samples. Thus, in samples with 3 examples a methodological error of standard deviation calculation is 23% and should be compensated.
Materials and methods. The article suggests to use imitation modeling means and to numerically obtain a density of standard deviation values distributin as a function of a number of examples in a training (test) sample. The work includes a table of values of multiplicative methodological errors of standard deviation calculations.
Results and conclusions. The authors have corrected the classical formulas for calculation of standard deviation and correlation coefficient taking into account compensation of methodological errors thereof due to a small number of examples in a test sample. The work displays a graph of real data on methodlogical errors and a graph of their analytical approximation by hyperbola.

Key words

methodological error, calculation of standard deviation on small samples, biometric data processing

Download PDF
References

1. Dodis Y., Reyzin L., Smith A. Fuzzy Extractors: How to Generate Strong Keys from Biometrics and Other Noisy. EUROCRYPT. 2004, April 13, pp. 523–540.
2. Monrose F., Reiter M., Li Q., Wetzel S. Cryptographic key generation from voice. Proc. IEEE Symp. on Security and Privacy, 2001.
3. Ramírez-Ruiz J., Pfeiffer C., Nolazco-Flores J. Cryptographic Keys Generation Using FingerCodes. Advances in Artificial Intelligence – IBERAMIA-SBIA 2006 (LNCS 4140). 2006, pp. 178–187.
4. Hao F., Anderson Ross and Daugman John Crypto with Biometrics Effectively. IEEE TRANSACTIONS ON COMPUTERS. 2006, September, vol. 55, no. 9.
5. Volchikhin V. I., Ivanov A. I., Funtikov V. A. Bystrye algoritmy obucheniya neyrosetevykh mekhanizmov biometriko-kriptograficheskoy zashchity informatsii: monogr. [Fast training lagorithms for neural-network mechanisms of biometric-cryptographic data protection: monograph]. Penza: Izd-vo PGU, 2005, 273 p.
6. 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.
7. GOST R 52633.5–2011. Zashchita informatsii. Tekhnika zashchity informatsii. Avtomaticheskoe obuchenie neyrosetevykh preobrazovateley biometriya-kod dostupa [Data protection. Data protection technology. Automatic training of neural network “biometrics-access code” converters]. Moscow, 2011.
8. Serikova N. I., Ivanov A. I., Kachalin S. V. Vestnik SibGAU [Bulletin of SSAU]. 2014, no. 3 (55), pp. 146–150.
9. Volchikhin V. I., Ivanov A. I., Serikova N. I., Funtikova Yu. V. Izvestiya vysshikh uchebnykh zavedeniy. Povolzhskiy region. Tekhnicheskie nauki [University proceedings. Volga region. Engineering sciences]. 2015, no. 2 (34), pp. 50–59.

 

Дата создания: 01.07.2016 09:12
Дата обновления: 01.07.2016 10:54