On biometric systems: electrocardiogram Gaussianity and data synthesis

Abstract

Electrocardiogram is a slow signal to acquire, and it is prone to noise. It can be inconvenient to collect large number of ECG heartbeats in order to train a reliable biometric system; hence, this issue might result in a small sample size phenomenon which occurs when the number of samples is much smaller than the number of observations to model. In this paper, we study ECG heartbeat Gaussianity and we generate synthesized data to increase the number of observations. Data synthesis, in this paper, is based on our hypothesis, which we support, that ECG heartbeats exhibit a multivariate normal distribution; therefore, one can generate ECG heartbeats from such distribution. This distribution is deviated from Gaussianity due to internal and external factors that change ECG morphology such as noise, diet, physical and psychological changes, and other factors, but we attempt to capture the underlying Gaussianity of the heartbeats. When this method was implemented for a biometric system and was examined on the University of Toronto database of 1012 subjects, an equal error rate (EER) of 6.71% was achieved in comparison to 9.35% to the same system but without data synthesis. Dimensionality reduction is widely examined in the problem of small sample size; however, our results suggest that using the proposed data synthesis outperformed several dimensionality reduction techniques by at least 3.21% in EER. With small sample size, classifier instability becomes a bigger issue and we used a parallel classifier scheme to reduce it. Each classifier in the parallel classifier is trained with the same genuine dataset but different imposter datasets. The parallel classifier has reduced predictors’ true acceptance rate instability from 6.52% standard deviation to 1.94% standard deviation.

Detecting Premature Ventricular Contraction by Using Regulated Discriminant Analysis with Very Sparse Training Data

“…However, a problem with this concept is low accuracy for small training set sizes (S Raudys 2015). Similar techniques are feature selection and feature extraction, which are very alike to dimensionality reduction techniques (Louis, et al 2017). A classifier which is based on a limited amount of training data is provided by Louis et al They assumed that the ECG signals are multivariate Gaussian distributions in a generative model which was used to generate training samples.…”

“…However, with small sample size they had instability problems which were solved by adding parallel classifiers that were trained with more data. It is noted that they used a proprietary database (Toronto database) for validating the scores (Louis, et al 2017). Andreao et al have used the Hidden Markov Model (HMM) to detect QRS complexes in a selected set from the MIT BIH database.…”

“…It is assumed that the distribution for the true ECG signals can be approximated by Gaussian distributions (Louis, et al 2017). The risk function uses the dirac-delta function (δ(p,q)), the unconditional prior (πi), the multivariate mean vector (μi), and the multivariate covariance matrix (Σi).…”

“…Unfortunately, estimators for covariance values are biased so large eigenvalues are biased towards higher values and small eigenvalues are biased towards lower values (Friedman 1988). Many approaches have been tried to remove this distortion from the eigenvalues and to make the covariance matrix nonsingular (Louis, et al 2017), (Murphy 2012). Especially in the context with small sample size settings a promising approach is deploying a regularization method that decreases the variance and regulate the covariance principal diagonal.…”

Deep Multiview Heartwave Authentication

“…This variation is one of the major dreaded challenges in inter and intra user authentication. The use of heartwave signal as biometric mode has aroused many research works with new and innovative approaches such as KNN classifiers (the most comment), LDA classifier [16], Support Vector Machine and Match Score Classifier [17,18] and Generative Model Classifier [19][20][21][22]. Unfortunately, all the works use ECG data that are obtained under resting condition.…”