English

应用科学学报 ›› 2022, Vol. 40 ›› Issue (1): 1-12.doi: 10.3969/j.issn.0255-8297.2022.01.001

• 计算机应用专辑 • 上一篇    下一篇

基于CNN和Bi-LSTM的脑电波情感分析

朱丽1,2,3, 杨青1,2,3, 吴涛1,2,3, 李晨1,2,3, 李铭1,2,3   

  1. 1. 华中师范大学 人工智能与智慧学习湖北省重点实验室, 湖北 武汉 430079;
    2. 华中师范大学 计算机学院, 湖北 武汉 430079;
    3. 华中师范大学 国家语言资源监测与研究网络媒体中心, 湖北 武汉 430079
  • 收稿日期:2021-07-13 发布日期:2022-01-28
  • 通信作者: 杨青,副教授,研究方向为深度学习和脑电波。E-mail:yangqing@mail.ccnu.edu.cn E-mail:yangqing@mail.ccnu.edu.cn
  • 基金资助:
    湖北省重点研发计划项目基金（No.2020BAB017）；武汉市科技计划项目基金（No.2019010701011392）；国家语委科研中心项目基金（No.ZDI135-135）资助

Emotional Analysis of Brain Waves Based on CNN and Bi-LSTM

ZHU Li1,2,3, YANG Qing1,2,3, WU Tao1,2,3, LI Chen1,2,3, LI Ming1,2,3   

  1. 1. Hubei Provincial Key Laboratory of Artificial Intelligence and Smart Learning, Central China Normal University, Wuhan 430079, Hubei, China;
    2. School of Computer, Central China Normal University, Wuhan 430079, Hubei, China;
    3. National Language Resources Monitoring & Research Center for Network Media, Central China Normal University, Wuhan 430079, Hubei, China
  • Received:2021-07-13 Published:2022-01-28

PDF

407

摘要： 针对目前大多数脑电波情感识别方法存在的依赖手动特征提取等问题，提出一种基于卷积神经网络（convolutional neural network，CNN）和双向长短时记忆（bidirectional long short-term memory，Bi-LSTM）网络的混合模型。首先将一维数据转换为二维数据，采用CNN提取空间特征；然后将一维数据输入Bi-LSTM，获取时间特征；最后将融合的空间和时间特征输入Softmax分类器，得到最终分类结果。在DEAP数据集上的实验结果表明：CNN和Bi-LSTM混合模型具有较好的分类性能，在效价度和唤醒度上的准确率分别达到88.55%和89.07%，是一种可行的脑电波情感分类模型。

关键词: 脑电信号, 情感分类, 卷积神经网络, 双向长短时记忆网络, 深度学习

Abstract: Aiming at the problem that most emotion recognition methods rely on manual feature extraction, a hybrid model based on convolutional neural network (CNN) and bidirectional long short-term memory (Bi-LSTM) network is proposed. Firstly, one-dimensional data is converted into two-dimensional data, and spatial features are extracted by CNN. Then the one-dimensional data is input into Bi-LSTM to obtain temporal features. Finally, the fused spatial and temporal features are input into Softmax classifier to obtain final classification results. Experimental results on DEAP dataset show that CNN and BiLSTM hybrid model has good classification performance, and the accuracy in potency and arousal reaches 88.55% and 89.07%, respectively, proving the proposed model is a feasible and affective EEG emotion classification model.

Key words: electroencephalogram, emotion classification, convolutional neural network (CNN), bidirectional long short-term memory (Bi-LSTM) network, deep learning

中图分类号: