Personal Identification by Convolutional Neural Network with ECG Signal

This paper proposes a novel approach for personal biometric identification with Convolutional Neural Network (CNN) based on the electrocardiogram (ECG) signals that are measured during bathing, aiming at exploring a feasible way to avoid or reduce the drowning accidents during daily bathing. After we perform denoising and QRS segmentation on the raw ECG signal, the preliminary experimental results showed that the best and robust identification rate is 97.20% for 10 individuals only with 5 epochs. When the proposed method was tested on a public ECG Database, the identification rate is as high as 98.70%. This work provides with strong evidence that ECG signals can be useful for personal identification.