基于深度置信网络的心电信号分类研究

Classification of ECG Signals Based on Deep Belief Network

提出一种基于深度置信网络(DBN)的心电信号分类方法。在对心电信号进行滤波等预处理后,构建一个DBN,由高斯-伯努利和伯努利-伯努利两种受限玻尔兹曼机(RBM)叠加而成。预训练阶段通过对比散度(CD)算法来学习RBM参数。权值微调阶段,在DBN顶部添加一个Softmax输出层,使用反向传播算法来执行有监督的微调。使用MIT-BIH心律失常数据库对DBN进行训练和测试,实验结果表明,和传统的人工设计方法相比,该方法在心电信号分类上性能更佳。

Proposes a classification method for ECG signals based on Deep Belief Network(DBN).After pre-processing the ECG signal,constructs a DBN to superimpose Gauss-Bernoulli and Bernoulli-Bernoulli restricted Boltzmann machines(RBM)to form DBN.At the pre-training phase,the parameters of RBM are learned by contrastive divergence training algorithms to initialize the DBN.In the fine-tuning phase,a Softmax output layer is added at the top of the DBN,and the backpropagation algorithm is used to perform supervised fine-tuning.The DBN is trained and tested on the MIT-BIH arrhythmia database.The experimental results show that compared with the traditional manual design method,the method has better performance in ECG signal classification.