Deep learning-based classification of massive electrocardiography data

Classification is the basis of electrocardiography (ECG) analysis. In the last decades, a large number of methods were proposed to deal with the classification of ECG beats. In this paper a kind of deep learning method is introduced into ECG beats classification. We create a classifier with stacked sparse autoencoder (SAE), and then combine the softmax regression with the SAE networks to consummate the classifier, from which we can get higher accuracy in the classification task. In the deep networks architecture, we use the stacked sparse autoencoder to get high-level features. Experimental results with the MIT-BIH ECG dataset confirmed that classifier build in this autoencoder based deep networks method perform better in the classical classification problem.