Publication

Abstract : Security now a days demanding highly advanced technologies to avoid falsification. There are solid evidences that the human heart beat can be used for biometric security. These trend in physiological signals used in this proposed work where deep learning methodologies incorporated for evaluation. The ECG-based biometric systems are usually less accurate than technologies based on other physiological traits. However ECG found out the best because of its uniqueness, specificity and unidimesional nature. In this work off-the-person category is considered and this mode of acquisition helps to minimize the complexity in acquiring the signal. The proposed work in this literature used raw ECG biometric data and convolutional neural network model for feature extraction and fully connected layers for classification. The performance comparison of different layers of Convolutional Neural Network (CNN) were compared with CNN-LSTM model. We have achieved the better performance in terms of accuracy with reduced Equal Error Rate(EER) with different CNN models.