A Novel Features Learning Method for ECG Arrhythmias Using Deep Belief Networks

In this paper, we propose a novel approach based on deep belief networks for features learning of ECG arrhythmias. The method consists of four steps: ECG signals preprocessing, segmentation and resampling, features learning, and validation. In the resampling stage all segmented ECG beats are changed into the same periodic length for the deep learning model. For extracting features of ECG data, we use two types of restricted boltzmann machine including Gaussian-Bernoulli restricted boltzmann machine and Bernoulli-Bernoulli restricted boltzmann machine to construct a deep belief netowrks. In order to enhance the deep belief netowrks performance, a fine-tuning process is carried out, which uses backpropagation by adding a softmax regression layer on the top of the resulting hidden representation layer to perform supervised classification. The experiment results show our approach achieves better performance with less feature learning time than traditional hand-designed methods on classification of ECG arrhythmias.