logo

SCIENTIA SINICA Informationis, Volume 48, Issue 7: 919-931(2018) https://doi.org/10.1360/N112017-00297

Research on EEG recognition based on improved-common spatial patterns and deep-belief network algorithm

More info
  • ReceivedMar 14, 2018
  • AcceptedApr 8, 2018
  • PublishedJul 16, 2018

Abstract

Using electroencephalogram (EEG) signals to control intelligent wheelchairs is one of the new control methods for controlling intelligent wheelchairs. The most serious problem in the control process is recognition and classification of EEG signals, especially a variety of EEG signals. To improve the accuracy of EEG classification, improved-common spatial patterns and a deep-belief network algorithm are proposed and used for recognition and classification of EEG signals. A variety of different EEG signals were collected by an Emotiv EPOC+ EEG collector, and the characteristic signals were extracted by improved-common spatial patterns and identified and classified by a deep-belief network algorithm. Simulation results show that the classification accuracy of improved-common spatial patterns and the deep-belief network algorithm is higher than that of the traditional classification method, and a research perspective is provided for the classification of various EEG signals in the future.


Funded by

北京信息科技大学重点研究培育项目(5221823307)

研究生科技创新项目(5121723303)

大学生创业培育基金项目(5111710813)


References

[1] Zhang R. A study on brain computer interface (BCI) — based functional assistance for the seriously disabled. Dissertation for Ph.D. Degree. Guangzhou: South China University of Technology, 2016. Google Scholar

[2] Hoffmann U, Vesin J M, Ebrahimi T. An efficient P300-based brain-computer interface for disabled subjects. J neurosci Method, 2008, 167: 115-125 CrossRef Google Scholar

[3] Li X. Study and implementation of online motor imagery based brain-computer interface system. Dissertation for Master Degree. Beijing: Tsinghua University, 2012. Google Scholar

[4] Wang S F, Lee Y H, Shiah Y J, et al. Time-frequency analysis of EEGs recorded during meditation. In: Proceedings of the 1st International Conference on Robot, Vision and Signal Processing (RVSP), Kaohsiung, 2011. 73--76. Google Scholar

[5] Wang K, Ye C, Shen Y Q, et al. Automatic removal algorithm of ocular artifact in electroencephalogram signal. Comput Eng, 2011, 37: 257--260. Google Scholar

[6] Rajesh P N Rao. Brain-computer Interfacing an Introduction. Beijing: China Machine Press, 2016. Google Scholar

[7] Gao S K. Comments on recent progress and challenges in the study of brain-computer interface. Chinese J Biomedical Eng, 2007, 26: 801--803. Google Scholar

[8] Dornhege G. Toward Brain-Computer Interfacing. Cambridge: MIT Press, 2007. Google Scholar

[9] Tang X L, Zhou J L, Zhang N, et al. Recognition of motor imagery EEG based on deep belief network. Inf Control, 2015, 44: 717--721. Google Scholar

[10] Ma M Z, Guo L B, Su K F. Classification method of motor imagery EEG signals based on wavelet and common spatial pattern. China Sci Technol Inf, 2017, 08: 83--85. Google Scholar

[11] Li L T. Classification method of four-class motor imagery EEG data based on common spatial pattern. Instrumentation, 2016, 05: 12--14. Google Scholar

[12] Jia X W, Li K, Li X Y, et al. A novel semi-supervised deep learning framework for affective state recognition on EEG signals. In: Proceedings of the 14th International Conference on Bioinformatics and Bioengineering, Boca Raton, 2014. Google Scholar

[13] Li X Y, Jia X W, Xun G X, et al. Improving EEG feature learning via synchronized facial video. In: Proceedings of International Conference on Big Data, Santa Clara, 2015. 843--848. Google Scholar

[14] Wang J W. Preprocessing methods and applications based on EEG. Dissertation for Master Degree. Beijing: Beijing University of Posts and Telecommunications, 2015. Google Scholar

[15] Liang J K. EEG Analysis and BCI Research based on Motor Imagery under Driving Behavior. Beijing: National Defense Industry Press, 2015. Google Scholar

[16] Zhang S H. Analysis of motor imagery EEG. Dissertation for Master Degree. Shanghai: Shanghai Jiao Tong University, 2015. Google Scholar

[17] Wang T. SMS classification methods based on deep learning. Dissertation for Master Degree. Xi'an: Chang'an University, 2016. Google Scholar

[18] Yang D P. Product image classification based on deep learning. Dissertation for Master Degree. Dalian: Dalian Jiaotong University, 2015. Google Scholar

[19] Dai R M. The motor imagery EEG classification based on deep learning. Dissertation for Master Degree. Beijing: Beijing Institute of Technology, 2015. Google Scholar

[20] Gao Y B, Lee H J, Mehmood R M, et al. Deep learning of EEG signals for emotion recognition. In: Proceedings of IEEE International Conference on Multimedia & Expo Workshops (ICMEW), Turin, 2016. Google Scholar

[21] Liu J W, Cheng Y, Zhang W D, et al. Deep learning EEG response representation for brain computer interface. In: Proceedings of the 34th Chinese Control Conference (CCC), Hangzhou, 2015. 3518--3523. Google Scholar

  • Figure 1

    Emotiv EPOC+ channel distribution of EEG

  • Figure 2

    An experimental process about three types of EEG

  • Figure 3

    (Color online) Motion (a) and standstill (b) EEG waveform of FC5 channel

  • Figure 4

    (Color online) Motion (a) and standstill (b) EEG waveform feature extraction by CSP

  • Figure 5

    (Color online) Restricted Boltzmann machine

  • Figure 6

    (Color online) Improved-common spatial patterns and deep belief network

  • Table 1   The correct recognition rate about first type EEG under different subjects in different classification methods
    DBN (%) SVM (%) (CSP+SVM) (%) (CSP+DBN) (%)
    Experimenter 1 88.33 78.33 85.83 93.33
    Experimenter 2 85.83 76.67 82.50 91.67
    Experimenter 3 86.67 82.50 83.33 94.17
    Experimenter 4 82.50 78.33 80.83 90.83
    Experimenter 5 91.67 84.17 87.50 96.67
    Experimenter 6 83.33 79.17 81.67 92.50
    Experimenter 7 81.67 75.83 80.83 89.17
    Experimenter 8 83.33 75.00 79.17 90.00
  • Table 2   The correct recognition rate about second type EEG under different subjects in different classification methods
    DBN (%) SVM (%) (CSP+SVM) (%) (CSP+DBN) (%)
    Experimenter 1 65.00 48.33 63.33 76.67
    Experimenter 2 61.67 43.33 58.33 73.33
    Experimenter 3 68.33 51.67 61.67 78.33
    Experimenter 4 68.33 48.33 58.33 75.00
    Experimenter 5 71.67 53.33 66.67 81.67
    Experimenter 6 63.33 45.00 55.00 71.67
    Experimenter 7 66.67 48.33 56.67 73.33
    Experimenter 8 63.33 51.67 60.00 76.67
  • Table 3   The correct recognition rate about third type EEG under different subjects in different classification methods
    DBN (%) SVM (%) (CSP+SVM) (%) (CSP+DBN) (%)
    Experimenter 1 40 25 35 47
    Experimenter 2 45 22 36 49
    Experimenter 3 47 28 40 51
    Experimenter 4 46 30 41 53
    Experimenter 5 49 27 41 52
    Experimenter 6 40 29 30 43
    Experimenter 7 43 31 36 48
    Experimenter 8 45 33 39 52

Copyright 2019 Science China Press Co., Ltd. 《中国科学》杂志社有限责任公司 版权所有

京ICP备18024590号-1