基于无阈值递归图和深度残差网络的脑电信号情感识别

提出基于无阈值递归图和深度残差网络相结合的脑电信号情感识别方法。基于非线性动力学理论,将脑电信号转化为无阈值递归图,克服了传统递归图分析中阈值选取的问题,同时脑电信号非线性特征被映射到二维平面。通过深度残差网络实现特征图非线性特征的自动提取,建立情感脑电分类模型,实现了单导联脑电信号情感识别。为进一步提高识别精度,联合四个单导联识别结果,采用“投票法”完成多导联脑电信号情感状态的联合识别。仿真结果表明,对Fp1、Fp2、F3、F4单导联脑电信号情感识别,平均准确率分别为93.82%、93.62%、94.54%、92.92%;多导联平均准确率为94.95%,提高了识别的准确率,具有很大的实用价值。

EEG EMOTIONAL RECOGNITION BASED ON THRESHOLDLESS RECURRENCE PLOT AND DEEP RESIDUAL NETWORK

This paper proposes an EEG emotion recognition method based on the combination of thresholdless recurrence plot and deep residual network.Based on the theory of nonlinear dynamics,the EEG signal was transformed into a thresholdless recurrence plot,which overcame the problem of threshold selection in traditional recurrence plot analysis.The nonlinear features of EEG signals were mapped to the two-dimensional plane.By using the deep residual network,automatic extraction of nonlinear features of feature maps was realized,emotional EEG classification model was established,and single-lead EEG emotion recognition was realized.To further improve recognition accuracy,combining with four single-lead recognition results,“voting method”was used to complete multi-lead joint recognition of the emotional state of the EEG signal.The simulation results show that the average accuracy of Fp1,Fp2,F3 and F4 single-lead EEG emotion recognition is 93.82%,93.62%,94.54%,92.92%,respectively;the average accuracy of multi-lead is 94.95%,which is improved.The accuracy of recognition has great practical value.