3차 텐서기반 MPCA 방법을 이용한 심전도신호의 개인식별

Abstract

In this paper, performance of individual identification on electrocardiogram using third-order tensor-based MPCA(Multilinear Principal Component Analysis) is performed. This method preserves the data structure by extracting features directly from the tensor representation without structural transformation of the data due to the vectorization process in order to reduce the dimension. It is also less susceptible to small data problems because it can learn more compact and potentially useful representation, and it can efficiently handle large tensors. Here, the third-order tensor is formed by reordering the one-dimensional electrocardiogram signal into a two-dimensional matrix and then taking the time frame into account. Physionet's PTB(Physikalisch-Technische Bundesanstalt) diagnostic database for performance evaluation is used, and MPCA showed 91.85% accuracy.