基于注意力机制RNN模型的癫痫患者脑电信号识别方法

EEG Recognition Method for Epileptic Patients Based on RNN Model with Attention Mechanism

针对癫痫患者脑电信号(electroencephalogram,EEG)数据识别提出了一种基于注意力机制的RNN(recurrent neural networks)模型.传统EEG特征分析耗时巨大且过度依赖专家经验,极大限制了脑活动识别方法的应用推广.因此,提出一种新的EEG识别方法以解决上述问题.首先对癫痫患者EEG的基本特征进行分析,进而采用基于注意力机制RNN模型消除各种干扰信号,利用XGBoost分类器识别EEG数据的类别,达到自动细化识别原始EEG的目的,最后在公共EEG数据集上进行大量实验,验证所提方法对EEG识别的准确性.实验结果表明,与一些成熟的EEG识别方法相比,本文所提方法在识别精度上有了进一步提升.

A RNN(recurrent neural networks)model based on attention mechanism is proposed for EEG(electroencephalogram)data recognition in epilepsy patients.Traditional EEG feature analysis is time-consuming and excessively dependent on expert experience,which greatly limits the application and popularization of brain activity recognition methods.A new EEG recognition method to solve the above problems is proposed.Firstly,the basic characteristics of EEG from epilepsy patients are analyzed.Then,the RNN model based on attention mechanism is designed to eliminate various interference signals and the XGBoost classifier is used to identify the categories of EEG data,so as to achieve the purpose of automatic refinement and recognition of the original EEG.Finally,a large number of experiments are carried out on the public EEG data set to verify the accuracy of the proposed method.The experimental results show that compared with the mature EEG recognition methods,the proposed method has higher recognition accuracy.