Key result
Deep learning neural networks provided higher accuracy for ECG-based user authentication compared to other classification techniques when sufficient training data was available.
Deep learning neural networks offer a highly accurate method for biometric authentication using ECG signals.
Traditional password based authentication has been proven inadequate and the use of biometrics have provided multiple solutions through the past years. One of the most recent approaches to biometric authentication is using Electrocardiograms (ECG), as they are closely related to unique characteristics of the heart of each person. In this paper a framework for efficient and usable user authentication, based on ECG, is proposed. The ECG is pre-processed in order to remove any noise or distortions and then multiple set of features are extracted from it, through various transformations. These set of features are used as input to classification models and the results are compared in order to find the most effective transformation-classifier combination, which also sets a performance baseline. Additionally Deep Learning Neural Networks are used in order to create classification models that predicts whether an ECG belongs to a specific person or not, based on the combination of the feature sets produced in the transformation step. The results obtained have shown that deep learning neural networks can provide higher accuracy in comparison to most of other techniques used, if there are enough data to train them on.
No takes yet. Share an insight, caveat, or question.
Chamatidis et al. (2017) studied Biometric authentication. Deep Learning Neural Networks vs. Other classification techniques was evaluated on Authentication accuracy. Deep learning neural networks provided higher accuracy for ECG-based user authentication compared to other classification techniques when sufficient training data was available.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: