基于降噪自编码神经网络的事件相关电位脑电信号分析方法
|
摘要
提出一种基于降噪自编码神经网络事件相关电位分析方法,首先建立3层神经网络结构,利用降噪自编码对神经网络进行初始化,实现了降噪自编码深度学习模型的无监督学习.从无标签数据中自动学习数据特征,通过优化模型训练得到的权值作为神经网络初始化参数.其次,经过有标签的样本进行网络参数的微调即可完成对神经网络的训练,该方法有效解决了神经网络训练中因随机选择初始化参数,而导致网络易陷入局部极小的缺陷.最后,利用上述神经网络对第3届脑机接口竞赛数据集Data set Ⅱ(事件相关电位脑电信号)进行分类分析.实验结果表明:利用降噪自编码迭代2500次训练神经网络模型,在受试者A和受试者B样本数据叠加5次、10次、15次3种情况下获得的分类准确率分别为73.4%, 87.4%和97.2%.该最高准确率优于其他分类方法,比竞赛第1名联合支持向量机(SVM)分类器(ESVM)提高了0.7%,为事件相关电位脑电信号提供了一种深度学习分析方法.
An algorithm based on denoising autoencoder neural network for event related potential analysis was proposed. Firstly, we establish a three layer neural network structure which is initialized by the denoising autoencoder. By using the unsupervised learning, the denoising autoencoder deep learning model is implemented. The weights obtained from the optimizing model are used as initialization parameters of the neural network by automatically learning of data characteristics from unlabeled data. Secondly, the training of the neural network can be completed through the fine-tuning of the network parameters with labeled data. This method effectively solves the problem of easy falling into local minimum for the neural network, which may be caused by random initialization. Thirdly, the proposed neural network was used in the competition Ⅲ Data set Ⅱ for classification analysis. Experimental results show that by using the denoising autoencoder neural network model under the training iterative of 2500, the average accuracy of 73.4%, 87.4% and 97.2%were obtained between subject A and subject B in three conditions which are the data is superimposed for 5, 10 and15 times respectively. This significant results show that our framework demonstrated superior performance in the higher classification than other methods(97.2% in comparison the highest accuracy 96.5%). In summary, we provided a denoising autoencoder neural network, which can learn more robust features from training data automatically. This deep learning model would be a new method for event-related potential electroencephalogram(EEG) signals analysis.
An algorithm based on denoising autoencoder neural network for event related potential analysis was proposed. Firstly, we establish a three layer neural network structure which is initialized by the denoising autoencoder. By using the unsupervised learning, the denoising autoencoder deep learning model is implemented. The weights obtained from the optimizing model are used as initialization parameters of the neural network by automatically learning of data characteristics from unlabeled data. Secondly, the training of the neural network can be completed through the fine-tuning of the network parameters with labeled data. This method effectively solves the problem of easy falling into local minimum for the neural network, which may be caused by random initialization. Thirdly, the proposed neural network was used in the competition Ⅲ Data set Ⅱ for classification analysis. Experimental results show that by using the denoising autoencoder neural network model under the training iterative of 2500, the average accuracy of 73.4%, 87.4% and 97.2%were obtained between subject A and subject B in three conditions which are the data is superimposed for 5, 10 and15 times respectively. This significant results show that our framework demonstrated superior performance in the higher classification than other methods(97.2% in comparison the highest accuracy 96.5%). In summary, we provided a denoising autoencoder neural network, which can learn more robust features from training data automatically. This deep learning model would be a new method for event-related potential electroencephalogram(EEG) signals analysis.
引文
[1]BASAR E,BASAR-EROGLU C,ROSEN B,et al.A new approach to endogenous event-related potentials in man:relation between EEGand P300-wave.International Journal of Neuroscience,1984,24(1):1-21.
[2]PINEGGER A,FALLER J,HALDER S,et al.Control or non-control state:that is the question an asynchronous visual P300-based BCI approach.Journal of Neural Engineering,2015,12(1):014001.
[3]WANG Hongtao,ZOU Heliang.Asynchronous TV remote control system based on event-related potential brain-computer interface.Control Theory&Applications,2012,29(11):1507-1511.(王洪涛,邹鹤良.基于事件关联电位脑机接口的电视遥控异步系统.控制理论与应用,2012,29(11):1507-1511.)
[4]WU B,SU Y,ZHANG J H,et al.A virtual Chinese keyboard BCI system based on P300 potentials.Acta Electronica Sinica,2009,37(8):1733-1745.
[5]DAL SENO B,MATTEUCCI M,MAINARDI L.Online detection of P300 and error potentials in a BCI speller.Computational Intelligence and Neuroscience,2010,10(3):11-15.
[6]GAREIS I E,ACEVEDO R C,ATUM Y V,et al.Determination of an optimal training strategy for a BCI classification task with LDA.The 5th International IEEE/EMBS Conference on IEEE Neural Engineering(NER).Cancun,Mexico:IEEE,2011:286-289.
[7]WANG H,LI Y,LONG J,et al.An asynchronous wheelchair control by hybrid EEG-EOG brain-computer interface.Cognitive Neurodynamics,2014,8(5):399-409.
[8]LECUN Y,BENGIO Y,HINTON G.Deep learning.Nature,2015,521(53):436-444.
[9]CECOTTI H,GRSER A.Convolutional neural networks for P300detection with application to brain-computer interfaces.IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(3):433-445.
[10]KULASINGHAM J P,VIBUJITHAN V,DE SILVA A C.Deep belief networks and stacked autoencoders for the P300 guilty knowledge test.2016 IEEE EMBS Conference on Biomedical Engineering and Sciences(IECBES).Ottawa,Canada:IEEE,2016:127-132.
[11]FERN′ANDEZ-REDONDO M,HERNANDEZ-ESPINOSA C.Weight initialization methods for multilayer feedforward.Proceedings-European Symposium on Artificial Neural Networks.Bruges,Belgium:D-Facto Public,2001,15(2):119-124.
[12]DENOEUX T,LENGELL′E R.Initializing back propagation networks with prototypes.Neural Networks,1993,6(93):351-363.
[13]SOCHER R,PENNINGTON J,HUANG E H,et al.Semi-supervised recursive autoencoders for predicting sentiment distributions.Proceedings of the Conference on Empirical Methods in Natural Language Processing.Portland,Oregon,USA:Association for Computational Linguistics,2011:151-161.
[14]ERHAN D,BENGIO Y,COURVILLE A,et al.Why does unsupervised pre-training help deep learning.Journal of Machine Learning Research,2010,11(3):625-660.
[15]GLOROT X,BORDES A,BENGIO Y.Domain adaptation for largescale sentiment classification:a deep learning approach.Proceedings of the Twenty-eight International Conference on Machine Learning.Bellevue,Washington,USA:Omnipress,2011:111-118.
[16]VINCENT P,LAROCHELLE H,LAJOIE I,et al.Stacked denoising autoencoders:learning useful representations in a deep network with a local denoising criterion.Journal of Machine Learning Research,2010,11(6):3371-3408.
[17]HU J.Comparison of different features and classifiers for driver fatigue detection based on a single EEG channel.Computational and Mathematical Methods in Medicine,2017,17(3):1-9.
[18]WANG Hongtao,LI Ting,HUANG Hui.A motor imagery analysis algorithm based on spatio-temporal-frequency joint selection and relevance vector machine.Control Theory&Applications,2017,34(10):1403-1408.(王洪涛,李霆,黄辉.一种基于时空频联合选择与相关向量机的运动想象脑电信号分析算法.控制理论与应用,2017,34(10):1403-1408.)
[19]ZONG R,ZHI Y,YAO B,et al.Classification and identification of soot source with principal component analysis and back-propagation neural network.Australian Journal of Forensic Sciences,2014,46(2):224-233.
[20]Bci competition III webpage,http://www.bbci.de/competition/iii/results/index.html.
[21]HALGREN E,MARINKOVIC K,CHAUVEL P.Generators of the late cognitive potentials in auditory and visual oddball tasks.Electroencephalography and Clinical Neurophysiology,1998,106(2):156-164.
[22]LIANG N,BOUGRAIN L.Averaging techniques for single-trial analysis of oddball event-related potentials.The 4th International Brain-Computer Interface Workshop.Autriche Graz:HAL-Inria,2008.
[23]ALAIN R,VINCENT G.BCI competition III:dataset II-ensemble of SVMs for BCI P300 speller.IEEE Transactions on Biomedical Engineering,2008,55(3):1147-1154.
[24]CHEA-YAU K,PONNAMBALAM S G,LOO C K.Multi-objective genetic algorithm as channel selection method for P300 and motor imagery data set.Neurocomputing,2015,161(1):120-131.
[2]PINEGGER A,FALLER J,HALDER S,et al.Control or non-control state:that is the question an asynchronous visual P300-based BCI approach.Journal of Neural Engineering,2015,12(1):014001.
[3]WANG Hongtao,ZOU Heliang.Asynchronous TV remote control system based on event-related potential brain-computer interface.Control Theory&Applications,2012,29(11):1507-1511.(王洪涛,邹鹤良.基于事件关联电位脑机接口的电视遥控异步系统.控制理论与应用,2012,29(11):1507-1511.)
[4]WU B,SU Y,ZHANG J H,et al.A virtual Chinese keyboard BCI system based on P300 potentials.Acta Electronica Sinica,2009,37(8):1733-1745.
[5]DAL SENO B,MATTEUCCI M,MAINARDI L.Online detection of P300 and error potentials in a BCI speller.Computational Intelligence and Neuroscience,2010,10(3):11-15.
[6]GAREIS I E,ACEVEDO R C,ATUM Y V,et al.Determination of an optimal training strategy for a BCI classification task with LDA.The 5th International IEEE/EMBS Conference on IEEE Neural Engineering(NER).Cancun,Mexico:IEEE,2011:286-289.
[7]WANG H,LI Y,LONG J,et al.An asynchronous wheelchair control by hybrid EEG-EOG brain-computer interface.Cognitive Neurodynamics,2014,8(5):399-409.
[8]LECUN Y,BENGIO Y,HINTON G.Deep learning.Nature,2015,521(53):436-444.
[9]CECOTTI H,GRSER A.Convolutional neural networks for P300detection with application to brain-computer interfaces.IEEE Transactions on Pattern Analysis and Machine Intelligence,2011,33(3):433-445.
[10]KULASINGHAM J P,VIBUJITHAN V,DE SILVA A C.Deep belief networks and stacked autoencoders for the P300 guilty knowledge test.2016 IEEE EMBS Conference on Biomedical Engineering and Sciences(IECBES).Ottawa,Canada:IEEE,2016:127-132.
[11]FERN′ANDEZ-REDONDO M,HERNANDEZ-ESPINOSA C.Weight initialization methods for multilayer feedforward.Proceedings-European Symposium on Artificial Neural Networks.Bruges,Belgium:D-Facto Public,2001,15(2):119-124.
[12]DENOEUX T,LENGELL′E R.Initializing back propagation networks with prototypes.Neural Networks,1993,6(93):351-363.
[13]SOCHER R,PENNINGTON J,HUANG E H,et al.Semi-supervised recursive autoencoders for predicting sentiment distributions.Proceedings of the Conference on Empirical Methods in Natural Language Processing.Portland,Oregon,USA:Association for Computational Linguistics,2011:151-161.
[14]ERHAN D,BENGIO Y,COURVILLE A,et al.Why does unsupervised pre-training help deep learning.Journal of Machine Learning Research,2010,11(3):625-660.
[15]GLOROT X,BORDES A,BENGIO Y.Domain adaptation for largescale sentiment classification:a deep learning approach.Proceedings of the Twenty-eight International Conference on Machine Learning.Bellevue,Washington,USA:Omnipress,2011:111-118.
[16]VINCENT P,LAROCHELLE H,LAJOIE I,et al.Stacked denoising autoencoders:learning useful representations in a deep network with a local denoising criterion.Journal of Machine Learning Research,2010,11(6):3371-3408.
[17]HU J.Comparison of different features and classifiers for driver fatigue detection based on a single EEG channel.Computational and Mathematical Methods in Medicine,2017,17(3):1-9.
[18]WANG Hongtao,LI Ting,HUANG Hui.A motor imagery analysis algorithm based on spatio-temporal-frequency joint selection and relevance vector machine.Control Theory&Applications,2017,34(10):1403-1408.(王洪涛,李霆,黄辉.一种基于时空频联合选择与相关向量机的运动想象脑电信号分析算法.控制理论与应用,2017,34(10):1403-1408.)
[19]ZONG R,ZHI Y,YAO B,et al.Classification and identification of soot source with principal component analysis and back-propagation neural network.Australian Journal of Forensic Sciences,2014,46(2):224-233.
[20]Bci competition III webpage,http://www.bbci.de/competition/iii/results/index.html.
[21]HALGREN E,MARINKOVIC K,CHAUVEL P.Generators of the late cognitive potentials in auditory and visual oddball tasks.Electroencephalography and Clinical Neurophysiology,1998,106(2):156-164.
[22]LIANG N,BOUGRAIN L.Averaging techniques for single-trial analysis of oddball event-related potentials.The 4th International Brain-Computer Interface Workshop.Autriche Graz:HAL-Inria,2008.
[23]ALAIN R,VINCENT G.BCI competition III:dataset II-ensemble of SVMs for BCI P300 speller.IEEE Transactions on Biomedical Engineering,2008,55(3):1147-1154.
[24]CHEA-YAU K,PONNAMBALAM S G,LOO C K.Multi-objective genetic algorithm as channel selection method for P300 and motor imagery data set.Neurocomputing,2015,161(1):120-131.
© 2004-2018 中国地质图书馆版权所有 京ICP备05064691号 京公网安备11010802017129号 地址：北京市海淀区学院路29号 邮编：100083 电话：办公室：(+86 10)66554848；文献借阅、咨询服务、科技查新：66554700 |