Sabodashko D. Improvement of methods and means of biometric authentication based on electrocardiogram

The paper solves an urgent scientific and applied problem in the field of cybersecurity - improving the characteristics of the biometric authentication system on the signal of the electrocardiogram based on a rational combination of digital signal processing and machine learning, which increases the security of resources at information facilities. Methods of ECG signal processing to increase the efficiency of biometric authentication systems have been developed and improved. An approach to the correction of residual artifacts in ECG signals has been developed. A study of the influence of hyperparameters on the accuracy of authentication. The method of forming the reference image of the ECGmarker on the basis of a statistical model, the method of forming the reference image of the ECGmarker on the basis of a fuzzy neural network model and the method of temporal normalization of heart rhythm are proposed.