基于时空Inception残差注意力网络的脑电情绪识别  

EEG emotion recognition based on spatiotemporal inceptionresidual attention network

作　　者：王伟[1] 周建华[1] 刘紫恒 赵世昊 伏云发[1] WANG Wei;ZHOU Jianhua;LIU Ziheng;ZHAO Shihao;FU Yunfa(School of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,P.R.China)

机构地区：[1]昆明理工大学信息工程与自动化学院,昆明650500

出　　处：《重庆邮电大学学报（自然科学版）》2024年第1期68-75,共8页Journal of Chongqing University of Posts and Telecommunications(Natural Science Edition)

基　　金：国家自然科学基金项目(82172058)~~。

摘　　要：为了提高脑电情绪识别分类精度,最大限度利用脑电信号的空间和时间信息,提出一种Inception残差注意力卷积神经网络与双向长短期记忆(bi-directional long short-term memory, BiLSTM)网络相结合的新型架构时空Inception残差注意力网络。将脑电信号采集电极位置映射到二维矩阵中,采集信号作为通道,构成三维数据;将得到的三维数据输入到时空Inception残差注意力卷积网络之中,提取时空信息;将得到的特征输入到全连接层进行分类;将Inception结构引入脑电情绪识别领域,实现多尺度特征提取,并将电极映射到矩阵之中,保留电极位置信息,使用时空Inception残差注意力网络从时空两个维度获取脑电相关信息。实验表明,使用该模型对DEAP数据集进行情绪四分类可得到93.71%的准确度,相较于对比模型,识别精度提高了10%~20%。提出的模型在脑电信号情绪识别领域具有优良性能。In order to improve the classification accuracy of electroencephalogram(EEG)emotion recognition and maximize the use of spatial and temporal information of EEG signals,a novel architecture spatio-temporal Inception residual attention convolutional neural network combined with bi-directional long short-term memory(BiLSTM)network is proposed.The electrode positions of EEG signals are mapped into a two-dimensional matrix,and the acquired signals are used as channels to form three-dimensional data;the obtained three-dimensional data are inputted into the spatio-temporal Inception residual attention convolutional network to extract the spatio-temporal information;and the obtained features are inputted into the fully connected layer for classification.In this paper,the Inception structure is introduced into the field of EEG emotion recognition,multi-scale feature extraction is realized,and the electrodes are mapped into the matrix,the electrode position information is retained,and the spatio-temporal Inception residual attention network is used to obtain the EEG-related information from the spatio-temporal dimension.Experiments show that 93.71%accuracy can be obtained by using the model for emotion IV classification of DEAP dataset,and the recognition accuracy is improved by 10%~20%compared with the comparison model.The proposed model has excellent performance in the field of EEG signal emotion recognition.

关 键 词：脑电信号 情绪识别 电极平面映射 Inception残差注意力网络 双向长短期记忆网络 

分 类 号：TN919[电子电信—通信与信息系统] TP391[电子电信—信息与通信工程]