基于回归神经网络的心电图分析

RECURRENT-NEURAL-NETWORK-BASED ECGANALYSIS

神经网络已成功地应用于文字识别和语音识别等领域，在心电信号的分析和处理中也越来越得到广泛的应用。本文研究了一种回归神经网络的结构及其学习算法，并在此基础上实现了基于回归神经网络的心电图动态分析系统。实验结果表明，该系统比基于ＢＰ神经网络的心电图分析系统具有更强的抑制噪声的能力，并能及时捕捉到心电图中动态变化的信息，从而进一步提高了心电图的分析效率。

In this paper, the structure and algorithm of recurrent neural net-work was studied, and recurrent-neural-network-based ECG dynamicanalysis system realized. The experiment results showed that this sy-stem was more robust against noise and could catch dynamic changesof ECG information, thus enhancing the efficiency of ECG analysis.