脑电信号特征的归一化方式与选择方法研究
|
摘要
针对多被试脑电数据存在被试间特征值差异较大的问题,分析了单次归一化的数据范围对分类准确率的影响。实验在情感数据集上采用6种常用的归一化方法,对所有被试的特征、单个被试的所有特征、单个被试的单个属性特征这三种单次归一化的数据范围进行准确率上的比较,证明了单个被试的单个属性特征更适于作为多被试脑电数据单次归一化数据范围。此外,提出方差贡献率与F-score结合的特征选择方法,在不降低准确率的情况下大量减少了特征数量。小波包树结点能量作为变换最少的特征得到的分类结果最好,小波包熵比脑电节律小波熵的分类准确率高。
Electroencephalographic( EEG) feature values of multiple subjects vary greatly between individual subjects. In light of this,we analyse the influence of data range of single normalisation on classification accuracy. This experiment uses six kinds of common normalisation methods on emotion dataset to compare the accuracies from three kinds of data range in regard to single normalisation. They are the features of all subjects,all features of single subject and the single property feature of single subject. The result proves that the single feature property of single subject is more appropriate as the data range of single normalisation of multiple subjects' EEG data. In addition,we present the feature selection method which combines the variance contribution rate with F-score,it brings a significant reduction in the number of features without reducing the classification accuracies. Wavelet packet tree node energy gets the best classification accuracies as the changed lest feature. The wavelet packet entropy has higher classification accuracies than the EEG rhythm wavelet entropy.
Electroencephalographic( EEG) feature values of multiple subjects vary greatly between individual subjects. In light of this,we analyse the influence of data range of single normalisation on classification accuracy. This experiment uses six kinds of common normalisation methods on emotion dataset to compare the accuracies from three kinds of data range in regard to single normalisation. They are the features of all subjects,all features of single subject and the single property feature of single subject. The result proves that the single feature property of single subject is more appropriate as the data range of single normalisation of multiple subjects' EEG data. In addition,we present the feature selection method which combines the variance contribution rate with F-score,it brings a significant reduction in the number of features without reducing the classification accuracies. Wavelet packet tree node energy gets the best classification accuracies as the changed lest feature. The wavelet packet entropy has higher classification accuracies than the EEG rhythm wavelet entropy.
引文
[1]肖汉光,蔡从中.特征向量的归一化比较性研究[J].计算机工程与应用,2009,45(22):117-119.
[2]Sugi T,Kawana F,Nakamura M.Automatic EEG arousal detection for sleep apnea syndrome[J].Biomedical Signal Processing and Control,2009,4(4):329-337.
[3]龙正吉,刘光远.用小波变换提取心电信号的情感特征[J].西南大学学报:自然科学版ISTIC,2011,33(3):109-113.
[4]刘冲,赵海滨,李春胜,等.基于频带能量归一化和SVM-RFE的ECo G分类[J].仪器仪表学报,2011,32(3):534-539.
[5]Koelstra S,Yazdani A,Soleymani M,et al.Single trial classification of EEG and peripheral physiological signals for recognition of emotions induced by music videos[M].Brain Informatics.Springer Berlin Heidelberg,2010:89-100.
[6]吴明,黄志华.ERPs信号特征提取及分类方法的研究[J].福州大学学报:自然科学版,2011,39(1):54-59.
[7]肖毅,陈善广,韩东旭.基于脑电节律的小波熵分析方法[J].科技导报,2011,29(16):39-44.
[8]曾宪伟,赵卫明,盛菊琴.小波包分解树结点与信号子空间频带的对应关系及其应用[J].地震学报,2008,30(1):90-96.
[9]郑常宝,段晓波,李晓明.用Matlab小波包求功率频带分解规律[J].电测与仪表,2010,47(6):22-24.
[10]Kubany A,Mhabary Z,Gontar V.EEG simulation by 2D interconnected chaotic oscillators[J].Chaos,Solitons&Fractals,2011,44(1):1-8.
[11]De Pascalis V,Cozzuto G,Caprara G V,et al.Relations among EEG-alpha asymmetry,BIS/BAS,and dispositional optimism[J].Biological Psychology,2013,94(1):198-209.
[12]Koelstra S,Muhl C,Soleymani M,et al.Deap:A database for emotion analysis;using physiological signals[J].Affective Computing,IEEE Transactions on,2012,3(1):18-31.
[13]Lin Y P,Wang C H,Jung T P,et al.EEG-based emotion recognition in music listening[J].Biomedical Engineering,IEEE Transactions on,2010,57(7):1798-1806.
[14]Nie D,Wang X W,Shi L C,et al.EEG-based emotion recognition during watching movies[C]//Neural Engineering(NER),2011 5th International IEEE/EMBS Conference on.IEEE,2011:667-670.
[15]Uusberg A,Uibo H,Kreegipuu K,et al.EEG alpha and cortical inhibition in affective attention[J].International Journal of Psychophysiology,2013,89(1):26-36.
[16]Vecchiato G,Toppi J,Astolfi L,et al.Spectral EEG frontal asymmetries correlate with the experienced pleasantness of TV commercial advertisements[J].Medical&Biological Engineering&Computing,2011,49(5):579-583.
[2]Sugi T,Kawana F,Nakamura M.Automatic EEG arousal detection for sleep apnea syndrome[J].Biomedical Signal Processing and Control,2009,4(4):329-337.
[3]龙正吉,刘光远.用小波变换提取心电信号的情感特征[J].西南大学学报:自然科学版ISTIC,2011,33(3):109-113.
[4]刘冲,赵海滨,李春胜,等.基于频带能量归一化和SVM-RFE的ECo G分类[J].仪器仪表学报,2011,32(3):534-539.
[5]Koelstra S,Yazdani A,Soleymani M,et al.Single trial classification of EEG and peripheral physiological signals for recognition of emotions induced by music videos[M].Brain Informatics.Springer Berlin Heidelberg,2010:89-100.
[6]吴明,黄志华.ERPs信号特征提取及分类方法的研究[J].福州大学学报:自然科学版,2011,39(1):54-59.
[7]肖毅,陈善广,韩东旭.基于脑电节律的小波熵分析方法[J].科技导报,2011,29(16):39-44.
[8]曾宪伟,赵卫明,盛菊琴.小波包分解树结点与信号子空间频带的对应关系及其应用[J].地震学报,2008,30(1):90-96.
[9]郑常宝,段晓波,李晓明.用Matlab小波包求功率频带分解规律[J].电测与仪表,2010,47(6):22-24.
[10]Kubany A,Mhabary Z,Gontar V.EEG simulation by 2D interconnected chaotic oscillators[J].Chaos,Solitons&Fractals,2011,44(1):1-8.
[11]De Pascalis V,Cozzuto G,Caprara G V,et al.Relations among EEG-alpha asymmetry,BIS/BAS,and dispositional optimism[J].Biological Psychology,2013,94(1):198-209.
[12]Koelstra S,Muhl C,Soleymani M,et al.Deap:A database for emotion analysis;using physiological signals[J].Affective Computing,IEEE Transactions on,2012,3(1):18-31.
[13]Lin Y P,Wang C H,Jung T P,et al.EEG-based emotion recognition in music listening[J].Biomedical Engineering,IEEE Transactions on,2010,57(7):1798-1806.
[14]Nie D,Wang X W,Shi L C,et al.EEG-based emotion recognition during watching movies[C]//Neural Engineering(NER),2011 5th International IEEE/EMBS Conference on.IEEE,2011:667-670.
[15]Uusberg A,Uibo H,Kreegipuu K,et al.EEG alpha and cortical inhibition in affective attention[J].International Journal of Psychophysiology,2013,89(1):26-36.
[16]Vecchiato G,Toppi J,Astolfi L,et al.Spectral EEG frontal asymmetries correlate with the experienced pleasantness of TV commercial advertisements[J].Medical&Biological Engineering&Computing,2011,49(5):579-583.
版权所有：© 2023 中国地质图书馆 中国地质调查局地学文献中心