基于卷积神经网络和支持向量机的事件相关电位识别方法

于鸿伟; 谢俊; 何柳诗; 杨育喆; 张焕卿; 徐光华

doi:10.7652/xjtuxb202112006

您当前的位置：

首页 >

文章列表页 >

基于卷积神经网络和支持向量机的事件相关电位识别方法

更新时间：2025-07-09

- 基于卷积神经网络和支持向量机的事件相关电位识别方法
- An Event-Related Potential Recognition Method Based on Convolutional Neural Network and Support Vector Machine
- 西安交通大学学报 2021年55卷第12期页码：47-54
- 作者机构：
  
  西安交通大学机械工程学院,西安,710049
- 作者简介：
- 基金信息：
  
  : 2021-04-18。作者简介: 于鸿伟(1997—),男,硕士生;谢俊(通信作者),男,副教授。基金项目: 人因工程重点实验室开放基金资助项目(SYFD061903K)
- DOI：10.7652/xjtuxb202112006
  中图分类号： TP23
- 网络首发：2021-12-10，
  
  纸质出版：2021
- 稿件说明：
移动端阅览
于鸿伟, 谢俊, 何柳诗, 等. 基于卷积神经网络和支持向量机的事件相关电位识别方法[J]. 西安交通大学学报, 2021,55(12):47-54.

An Event-Related Potential Recognition Method Based on Convolutional Neural Network and Support Vector Machine[J]. 2021, 55(12): 47-54.
于鸿伟, 谢俊, 何柳诗, 等. 基于卷积神经网络和支持向量机的事件相关电位识别方法[J]. 西安交通大学学报, 2021,55(12):47-54. DOI： 10.7652/xjtuxb202112006.

An Event-Related Potential Recognition Method Based on Convolutional Neural Network and Support Vector Machine[J]. 2021, 55(12): 47-54. DOI： 10.7652/xjtuxb202112006.

摘要

针对事件相关电位(ERP)脑电信号存在个体差异性强、信噪比低等特点而导致其识别困难以及传统的卷积神经网络(CNN)对小样本分类易产生过拟合等问题

在CNN和支持向量机(SVM)融合模型的基础上

提出一种用于ERP信号分类识别的CNN-SVM组合分类器方法。该方法以滤波后的原始多通道ERP信号作为输入

首先使用一维时间卷积核对信号的时域进行卷积操作; 然后使用一维空间卷积核进行空堿卷积

从信号的时间信息和空间信息中学习特征; 最后通过降采样、全连接等操作完成CNN模型的训练; 接着再次将信号导入训练好的CNN模型来提取信号的降采样层特征

最后利用SVM对信号降采样层特征进行分类识别。实验结果表明:利用该组合分类器方法可以在少量重复的视觉刺激下实现对受试者P300信号的有效识别; 在重复4次以上刺激后

该组合分类器的平均识别准确率为93.49%

相比于传统的CNN方法

其平均识别准确率提升了4.36%; 与用于ERP信号识别的经典算法——逐步线性判别分析和贝叶斯线性判别分析相比

该组合分类器的平均识别正确率分别提高了6.38%和4.16%。该组合分类器只需要少量的重复实验便可获得较高的目标识别准确率

有效提高了ERP信号的识别效果。

Abstract

In view of the difficulties in identifying the event-related potential(ERP)signals due to their characteristics of strong individual differences

and the overfitting problem of traditional convolutional neural network(CNN)for small samples

based on a fusion model of CNN and support vector machine(SVM)

a CNN and SVM combined classifier for ERP signal classification and recognition is proposed. This method takes the filtered original multi-channel ERP signals as input. Firstly

a one-dimensional time convolution kernel is used to convolve the time domain of the signals

then the spatial convolution is performed by using a one-dimensional spatial convolution kernel to learn features from temporal and spatial information of signals. Finally

the training of the CNN model is accomplished by such operations as down-sampling and full connection. After that

the signals are imported into the trained model again to extract the down-sampling layer features of the signals

and SVM is finally used to classify and identify the features. Classification results show that the proposed combined classifier can effectively recognize the P300 signal component under a small number of repeated visual stimuli. The average recognition accuracy of the proposed method is 94.08% after the repetition of more than four times of visual stimuli

and it has an average accuracy improvement of 4.36% compared with the traditional CNN method. Compared with the classical algorithms of stepwise linear discriminant analysis(SWLDA)and Bayesian linear discriminant analysis(BLDA)

the average recognition accuracy of the combined classifier is improved by 6.83% and 4.16%

respectively. The combined classifier only needs a small number of repeated experiments to obtain higher target recognition accuracy

and effectively improves the recognition effect of ERP signals.

关键词

Keywords

references

MCFARLAND D J, WOLPAW J R. Brain-computer interfaces for communication and control [J]. Communications of the ACM, 2011, 54(5): 60-66.

WANG Yijun, NAKANISHI M, ZHANG Dan. EEG-based brain-computer interfaces [J]. Advances in Experimental Medicine and Biology, 2019, 1101: 41-65.

尚淑怡, 尤春景. 认知电位P300的应用及研究进展 [J]. 中国康复, 2008(2): 33-135.

SHANG Shuyi, YOU Chunjing. Application and research progress of cognitive potential P300 [J]. China Rehabilitation, 2008(2): 133-135.

徐光华, 张锋, 谢俊, 等. 稳态视觉诱发电位的脑机接口范式及其信号处理方法研究 [J]. 西安交通大学学报, 2015, 49(6): 1-7.

XU Guanghua, ZHANG Feng, XIE Jun, et al. Brain computer interface paradigms and signal processing strategy for steady state visual evoked potential [J]. Journal of Xi'an Jiaotong University, 2015, 49(6): 1-7.

陈泽龙, 康宁. 基于脑电EEG信号的分析分类方法 [J]. 中国医学装备, 2019, 16(12): 151-158.

CHEN Zelong, XIE Kangning. Analysis and classification method based on EEG signal [J]. Chinese Medical Equipment, 2019, 16(12): 151-158.

KUNDU S, ARI S. P300 based character recognition using convolutional neural network and support vector machine [J]. Biomedical Signal Processing and Control, 2020, 55: 1-7.

MA Ye, JIANG Guangping, CHANG Tanqing, et al. Research on feature extraction and classification of P300 EEG signals [EB/OL]. [2021-01-10]. https:∥www.atlantis-press.com/proceedings/eame-17/25875 750.

KAPER M, MEINICKE P, LINGNER T, et al. Support vector machines for the P300 speller paradigm [J]. IEEE Transactions on Biomedical Engineering, 2004, 51(6): 1073-1076.

ORALHAN Z. 2 Stages-region-based P300 speller in brain-computer interface [J]. IEEE Journal of Research, 2019, 65(6): 740-748.

LABIB E, FOUAD I A, MAI S M, et al. Multiple classify-cation techniques toward a robust and reliable P300 BCI system [J]. Biomedical Engineering: Applications, Basis and Communications, 2020, 32(2): 63-71.

曹帅, 吴畏. 基于xDAWN和降噪自编码器的事件相关电位分类 [J]. 中国科技论文, 2018, 13(24): 2829-2833.

CAO Shuai, WU Wei. Event related potential classifi-cation based on xDAWN and noise reduction self-encoder [J]. Chinese Scientific Paper, 2018, 13(24): 2829-2833.

WANG Z, HEALY G, SMEATON A F, et al. Spatial filtering pipeline evaluation of cortically coupled computer vision system for rapid serial visual presentation [J]. Brain Computer Interfaces, 2018, 5(4): 132-145.

杜光景, 谢俊, 张玉彬, 等. 用于稳态视觉诱发电位脑机接口目标识别的深度学习方法 [J]. 西安交通大学学报, 2019, 53(11): 42-48.

DU Guangjing, XIE Jun, ZHANG Yubin, et al. A deep learning method for target recognition in steady-state visual evoked potential-based brain-computer interface [J]. Journal of Xi'an Jiaotong University, 2019, 53(11): 42-48.

CECOTTI H, GRASER A. Convolutional neural net-works for P300 detection with application to brain computer interfaces [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(3): 433-445.

LIU Mingfei, WU Wei, GU Zhenghui, et al. Deep learning based on batch normalization for P300 signal detection [J]. Neurocomputing, 2017, 275: 288-297.

SZEGEDY C, VANHOUCKE V, IOFFE S, et al. Rethinking the inception architecture for computer vision [C]∥Proceedings of the 29th IEEE Conference on Computer Vision and Pattern Recognition. Piscataway, NJ, USA: IEEE, 2016: 2818-2826.

黄孝喜, 李晗雨, 王荣波, 等. 基于卷积神经网络与SVM分类器的隐喻识别 [J]. 数据分析与知识发现, 2018, 2(10): 77-83.

HUANG Xiaoxi, LI Hanyu, WANG Rongbo, et al. Metaphor recognition based on convolutional neural network and SVM classifier [J]. Data Analysis and Knowledge Discovery, 2018, 2(10): 77-83.

LECUN Y, BOTTOU L, BENGIO Y, et al. Gradient-based learning applied to document recognition [J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.

SAMUEL S, PATRICIA T, JOSEPH Z, et al. Infor-mation delivery and the sensory evoked potential [J]. Science, 1967, 155(3768): 1436-1439.

BLANKERTZ B. BCI competition III webpage [EB/OL]. [2021-01-10]. http: ∥www.bbci.de/competition/iii/desc_II.pdf.

BLANKERTZ B. BCI competition II webpage [EB/OL]. [2020-12-15]. http: ∥www.bbci.de/competition/ii/albany_desc/abbany_desc_ii.pdf.

CLERC M. Brain computer interfaces, principles and practise [J]. Biomedical Engineering Online, 2013, 12(1): 1-4.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

蜉蝣优化双通道网络在齿轮箱故障诊断中的应用

一种适用于混合三端直流输电线路的故障定位方法

采用格拉姆角场-卷积神经网络-时序卷积网络混合模型的锂离子电池健康状态估计

面向分类去噪问题的模糊支持向量机新算法

利用Fisher判别式和事件相关电位的心理意识真实性识别