NEURAL NETWORK CONVERSION OF BIOMETRY INTO AUTHENTICATION CODE: ADDITION OF HAMMING ENTROPY WITH ENTROPY OF CORRELATION RELATIONS BETWEEN DISCHARGES

Abstract

Background. The problem of calculating the entropy of 256-bit codes with dependent bits on small test samples consisting of 20 examples is considered. Materials and methods. It is proposed to estimate the entropy of the output codes of the neural network transformer by calculating the mutual correlation coefficients of the code sequences 256 bits long, obtained for examples of one “Alien” image. Results. It is shown that the proposed method is much more accurate than the previously used estimation method by calculating the mathematical expectation and standard deviation of Hamming distances for the same “Alien” image. Conclusions. The results obtained make it possible to raise the question of adjusting the national standard GOST R 52633.3 in the near future through the introduction of an additional section into it concerning the calculation of the correlation entropy.

Keywords

• small sample testing
• artificial neurons
• biometrics-to-code conversion