一种基于深度学习的心电信号分类方法

An ECG Signal Classification Method Based on Deep Learning

提出了一种基于深度学习人工神经网络的心电信号分类方法.方法直接从原始数据抽象网络各层次上的基底特征,取代了基本波形定位和特征的人为选定.隐含层节点数逐级下降和深度学习的多隐含层深架构的采用,有效避免了已有方法中的定域性问题和网络规模.基于MIT-BIH数据库的实验表明,与传统BP网络和蚁群算法的分类方法相比,该方法总分类精度达93.3%.

An ECG signal classification method based on deep learning artificial neural network is proposed in this paper. Instead of manual selection of basic wave form positioning and features, this method directly abstracts the base features at each level of the network from the original data. A step-by-step reduction of the hidden layer nodes and multi-hidden layer architecture of the deep learning makes the new method effectively avoid the localization and large network scale in the existing methods. Experiments based on the MIT-BIH database show that, compared with the traditional BP network and ant colony algorithm classification methods, the proposed method can obtain better performance.