ELEKTRA: ELEKTRokardiomatrix application to biometric identification with convolutional neural networks

Biometric systems are an uprising technique of identification in today’s world. Many different have been used everyone’s daily life the past years, such as fingerprint, face scan, and others. We propose a new method using Elektrokardiogramms (EKGs) converted into heatmap set aligned R-peaks (heartbeats), forming matrix called Elektrokardiomatrix (EKM). can build one-against-many system Convolutional Neural Network (CNN). tested our proposal with one main database (the Normal Sinus Rhythm Database (NSRDB)) two other databases, which MIT-BIH Arrhythmia (MIT-BIHDB) Physikalisch-Technische Bundesanstalt (PTB) Database. With NSRDB, we achieved accuracy 99.53% offered False Acceptance Rate (FAR) 0.02% Rejection (FRR) 0.05%. Very similar results were also obtained PTB databases. performed in-depth experimentation to test efficiency feasibility novel biometric solution. It is remarkable that simple CNN, has only convolutional layer, max-pooling operation, some regularisation, identify users very high performance low error rates. Consequently, model does not need complex architectures offer high-performance metrics.