共空间模式结合小波包分解的脑电情感分类
|
摘要
为了有效缓解不同受试者跨天试验间脑电信号差异对分类性能的影响,结合共空间模式和小波包分解算法,对12个受试者连续5天的脑电波数据进行空间滤波处理和时频域上小波包能量特征提取。采用Bagging tree、SVM、LDA和BLDA模型进行情感分类实验。实验结果表明,使用SVM和BLDA分类器对该算法提取的脑电特征进行两类情感分类的精度比目前最优的结果分别提高了4.4%和3.5%,有效地提高了跨天脑电情感分类的准确率和稳定性,对于开发鲁棒的情感脑-机接口应用具有一定价值。
To effectively alleviate the effect of Electroencephalogram(EEG)differences on classification performance in different subjects during cross-day sessions, this paper suggests to combine Common Spatial Pattern(CSP)and Wavelet Packet Decomposition(WPD)algorithm to make filtering preprocessing and feature extraction on EEG data of 12 subjects for 5 consecutive days. The Bagging tree, SVM, LDA and BLDA models are used for the binary(happy and sad)emotional classification experiments. The experimental results show that the best classification accuracy of SVM and BLDA models on two types of EEG features extracted by CSP and WPD algorithms is respectively 4.4% and 3.5% higher than the stateof-the-art, which indicates that the proposed method can effectively improve the accuracy and stability of cross-day EEG emotion classification and has certain value for the development of robust affective brain-computer interface applications.
To effectively alleviate the effect of Electroencephalogram(EEG)differences on classification performance in different subjects during cross-day sessions, this paper suggests to combine Common Spatial Pattern(CSP)and Wavelet Packet Decomposition(WPD)algorithm to make filtering preprocessing and feature extraction on EEG data of 12 subjects for 5 consecutive days. The Bagging tree, SVM, LDA and BLDA models are used for the binary(happy and sad)emotional classification experiments. The experimental results show that the best classification accuracy of SVM and BLDA models on two types of EEG features extracted by CSP and WPD algorithms is respectively 4.4% and 3.5% higher than the stateof-the-art, which indicates that the proposed method can effectively improve the accuracy and stability of cross-day EEG emotion classification and has certain value for the development of robust affective brain-computer interface applications.
引文
[1]蒋静芳,曾颖,林志敏,等.基于脑电信号的情绪评估研究综述[J].信息工程大学学报,2016,17(6):686-693.
[2]陆苗,邹俊忠,张见,等.基于IMF能量熵的脑电情感特征提取研究[J].生物医学工程研究,2016,35(2):71-74.
[3] Kashihara K.A brain-cosmputer interface for potential non-verbal facial communication based on EEG signals related to specific emotions[J].European Journal of Clinical Investigation,2014,29(5):395-403.
[4] TsiourisΚΜ,Markoula S,Konitsiotis S,et al.A robust unsupervised epileptic seizure detection methodology to accelerate large EEG database evaluation[J].Biomedical Signal Processing and Control,2018,40:275-285.
[5] Lin Y P,Jung T P.Exploring day-to-day variability in EEG-based emotion classification[C]//Proceedings of IEEE International Conference on Systems,Man and Cybernetics,San Diego,CA,USA,2014:2226-2229.
[6] Jao P K,Lin Y P,Yang Y H,et al.Using robust principal component analysis to alleviate day-to-day variability in EEG based emotion classification[C]//Proceedings of Engineering in Medicine and Biology Society,Milan,Italy,2015:570-573.
[7] Lin Y P,Jao P K,Yang Y H.Improving cross-day eegbased emotion classification using robust principal component analysis[J].Frontiers in Computational Neuroscience,2017,64:1-11.
[8]张学军,黄婉露,黄丽亚,等.公共空间模式算法结合经验模式分解的EEG特征提取[J].计算机工程与应用,2017,53(13):9-15.
[9]周蚌艳,吴小培,吕钊,等.基于共空间模式方法的多类运动想象脑电的导联选择[J].生物医学工程学杂志,2015,32(3):520-525.
[10]吴林彦,鲁昊,高诺,等.基于CSP算法与小波包分析方法的运动想象脑电信号特征提取性能的比较[J].生物医学工程研究,2017,36(3):224-228.
[11]李昕,孙小棋,王欣,等.基于自回归小波包熵特征融合算法的情感识别研究[J].生物医学工程学杂志,2017,34(6):831-836.
[12]李明爱,张梦,孙炎珺.基于小波包和深度信念网络的脑电特征提取方法[J].电子测量与仪器学报,2018,32(1):111-118.
[13]张健钊,姜威,贲晛烨.基于样本熵与小波包能量特征提取和Real AdaBoost算法的正常期、癫痫间歇与发作期的脑电自动检测[J].生物医学工程学杂志,2016,33(6):1031-1038.
[14] Cheng Minmin,Lu Zuhong,Wang Haixian.Regularized common spatial patterns with subject-to-subject transfer of EEG signals[J].Cognitive Neurodynamics,2017,11(2):173-181.
[15] Song Xiaomu,Yoon Sukchung.Improving brain-computer interface classification using adaptive common spatial patterns[J].Computers in Biology and Medicine,2015,61:150-160.
[16] Ang Q X,Yi Q Y,Wee W.Emotion classification from EEG signals using time-frequency-DWT features and ANN[J].Journal of Computer&Communications,2017,5(3):75-79.
[17] Atkinson J,Campos D.Improving BCI-based emotion recognition by combining EEG feature selection and kernel classifiers[J].Expert Systems with Applications,2016,47:35-41.
[18] Zhuang N,Zeng Y,Tong L,et al.Emotion recognition from EEG signals using multidimensional information in EMD domain[J].BioMed Research International,2017(1):8317357.
[2]陆苗,邹俊忠,张见,等.基于IMF能量熵的脑电情感特征提取研究[J].生物医学工程研究,2016,35(2):71-74.
[3] Kashihara K.A brain-cosmputer interface for potential non-verbal facial communication based on EEG signals related to specific emotions[J].European Journal of Clinical Investigation,2014,29(5):395-403.
[4] TsiourisΚΜ,Markoula S,Konitsiotis S,et al.A robust unsupervised epileptic seizure detection methodology to accelerate large EEG database evaluation[J].Biomedical Signal Processing and Control,2018,40:275-285.
[5] Lin Y P,Jung T P.Exploring day-to-day variability in EEG-based emotion classification[C]//Proceedings of IEEE International Conference on Systems,Man and Cybernetics,San Diego,CA,USA,2014:2226-2229.
[6] Jao P K,Lin Y P,Yang Y H,et al.Using robust principal component analysis to alleviate day-to-day variability in EEG based emotion classification[C]//Proceedings of Engineering in Medicine and Biology Society,Milan,Italy,2015:570-573.
[7] Lin Y P,Jao P K,Yang Y H.Improving cross-day eegbased emotion classification using robust principal component analysis[J].Frontiers in Computational Neuroscience,2017,64:1-11.
[8]张学军,黄婉露,黄丽亚,等.公共空间模式算法结合经验模式分解的EEG特征提取[J].计算机工程与应用,2017,53(13):9-15.
[9]周蚌艳,吴小培,吕钊,等.基于共空间模式方法的多类运动想象脑电的导联选择[J].生物医学工程学杂志,2015,32(3):520-525.
[10]吴林彦,鲁昊,高诺,等.基于CSP算法与小波包分析方法的运动想象脑电信号特征提取性能的比较[J].生物医学工程研究,2017,36(3):224-228.
[11]李昕,孙小棋,王欣,等.基于自回归小波包熵特征融合算法的情感识别研究[J].生物医学工程学杂志,2017,34(6):831-836.
[12]李明爱,张梦,孙炎珺.基于小波包和深度信念网络的脑电特征提取方法[J].电子测量与仪器学报,2018,32(1):111-118.
[13]张健钊,姜威,贲晛烨.基于样本熵与小波包能量特征提取和Real AdaBoost算法的正常期、癫痫间歇与发作期的脑电自动检测[J].生物医学工程学杂志,2016,33(6):1031-1038.
[14] Cheng Minmin,Lu Zuhong,Wang Haixian.Regularized common spatial patterns with subject-to-subject transfer of EEG signals[J].Cognitive Neurodynamics,2017,11(2):173-181.
[15] Song Xiaomu,Yoon Sukchung.Improving brain-computer interface classification using adaptive common spatial patterns[J].Computers in Biology and Medicine,2015,61:150-160.
[16] Ang Q X,Yi Q Y,Wee W.Emotion classification from EEG signals using time-frequency-DWT features and ANN[J].Journal of Computer&Communications,2017,5(3):75-79.
[17] Atkinson J,Campos D.Improving BCI-based emotion recognition by combining EEG feature selection and kernel classifiers[J].Expert Systems with Applications,2016,47:35-41.
[18] Zhuang N,Zeng Y,Tong L,et al.Emotion recognition from EEG signals using multidimensional information in EMD domain[J].BioMed Research International,2017(1):8317357.