BWGAN-GP: An EEG Data Generation Method for Class Imbalance Problem in RSVP Tasks
In the rapid serial visual presentation (RSVP) classification task, the data from the target and non-target classes are incredibly imbalanced. These class imbalance problems (CIPs) can hinder the classifier from achieving better performance, especially in deep learning. This paper proposed a novel d...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2022-01-01
|
| Series: | IEEE Transactions on Neural Systems and Rehabilitation Engineering |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/9690467/ |
| _version_ | 1797805084642377728 |
|---|---|
| author | Meng Xu Yuanfang Chen Yijun Wang Dan Wang Zehua Liu Lijian Zhang |
| author_facet | Meng Xu Yuanfang Chen Yijun Wang Dan Wang Zehua Liu Lijian Zhang |
| author_sort | Meng Xu |
| collection | DOAJ |
| description | In the rapid serial visual presentation (RSVP) classification task, the data from the target and non-target classes are incredibly imbalanced. These class imbalance problems (CIPs) can hinder the classifier from achieving better performance, especially in deep learning. This paper proposed a novel data augmentation method called balanced Wasserstein generative adversarial network with gradient penalty (BWGAN-GP) to generate RSVP minority class data. The model learned useful features from majority classes and used them to generate minority-class artificial EEG data. It combines generative adversarial network (GAN) with autoencoder initialization strategy enables this method to learn an accurate class-conditioning in the latent space to drive the generation process towards the minority class. We used RSVP datasets from nine subjects to evaluate the classification performance of our proposed generated model and compare them with those of other methods. The average AUC obtained with BWGAN-GP on EEGNet was 94.43%, an increase of 3.7% over the original data. We also used different amounts of original data to investigate the effect of the generated EEG data on the calibration phase. Only 60% of original data were needed to achieve acceptable classification performance. These results show that the BWGAN-GP could effectively alleviate CIPs in the RSVP task and obtain the best performance when the two classes of data are balanced. The findings suggest that data augmentation techniques could generate artificial EEG to reduce calibration time in other brain-computer interfaces (BCI) paradigms similar to RSVP. |
| first_indexed | 2024-03-13T05:46:54Z |
| format | Article |
| id | doaj.art-5850255c14674368862342624e6872fe |
| institution | Directory Open Access Journal |
| issn | 1558-0210 |
| language | English |
| last_indexed | 2024-03-13T05:46:54Z |
| publishDate | 2022-01-01 |
| publisher | IEEE |
| record_format | Article |
| series | IEEE Transactions on Neural Systems and Rehabilitation Engineering |
| spelling | doaj.art-5850255c14674368862342624e6872fe2023-06-13T20:08:36ZengIEEEIEEE Transactions on Neural Systems and Rehabilitation Engineering1558-02102022-01-013025126310.1109/TNSRE.2022.31455159690467BWGAN-GP: An EEG Data Generation Method for Class Imbalance Problem in RSVP TasksMeng Xu0Yuanfang Chen1https://orcid.org/0000-0002-9888-9869Yijun Wang2https://orcid.org/0000-0001-9950-6025Dan Wang3https://orcid.org/0000-0002-0657-9370Zehua Liu4https://orcid.org/0000-0001-6544-4858Lijian Zhang5https://orcid.org/0000-0002-5953-0079Faculty of Information Technology, Beijing University of Technology, Beijing, ChinaBeijing Institute of Mechanical Equipment, Beijing, ChinaState Key Laboratory on Integrated Optoelectronics, Institute of Semiconductors, Chinese Academy of Sciences, Beijing, ChinaFaculty of Information Technology, Beijing University of Technology, Beijing, ChinaFan GongXiu Honors College, Beijing University of Technology, Beijing, ChinaFaculty of Information Technology, Beijing University of Technology, Beijing, ChinaIn the rapid serial visual presentation (RSVP) classification task, the data from the target and non-target classes are incredibly imbalanced. These class imbalance problems (CIPs) can hinder the classifier from achieving better performance, especially in deep learning. This paper proposed a novel data augmentation method called balanced Wasserstein generative adversarial network with gradient penalty (BWGAN-GP) to generate RSVP minority class data. The model learned useful features from majority classes and used them to generate minority-class artificial EEG data. It combines generative adversarial network (GAN) with autoencoder initialization strategy enables this method to learn an accurate class-conditioning in the latent space to drive the generation process towards the minority class. We used RSVP datasets from nine subjects to evaluate the classification performance of our proposed generated model and compare them with those of other methods. The average AUC obtained with BWGAN-GP on EEGNet was 94.43%, an increase of 3.7% over the original data. We also used different amounts of original data to investigate the effect of the generated EEG data on the calibration phase. Only 60% of original data were needed to achieve acceptable classification performance. These results show that the BWGAN-GP could effectively alleviate CIPs in the RSVP task and obtain the best performance when the two classes of data are balanced. The findings suggest that data augmentation techniques could generate artificial EEG to reduce calibration time in other brain-computer interfaces (BCI) paradigms similar to RSVP.https://ieeexplore.ieee.org/document/9690467/Rapid serial visual presentation (RSVP)Wasserstein generative adversarial network (WGAN)data augmentationclass imbalance problemauto-encoder |
| spellingShingle | Meng Xu Yuanfang Chen Yijun Wang Dan Wang Zehua Liu Lijian Zhang BWGAN-GP: An EEG Data Generation Method for Class Imbalance Problem in RSVP Tasks IEEE Transactions on Neural Systems and Rehabilitation Engineering Rapid serial visual presentation (RSVP) Wasserstein generative adversarial network (WGAN) data augmentation class imbalance problem auto-encoder |
| title | BWGAN-GP: An EEG Data Generation Method for Class Imbalance Problem in RSVP Tasks |
| title_full | BWGAN-GP: An EEG Data Generation Method for Class Imbalance Problem in RSVP Tasks |
| title_fullStr | BWGAN-GP: An EEG Data Generation Method for Class Imbalance Problem in RSVP Tasks |
| title_full_unstemmed | BWGAN-GP: An EEG Data Generation Method for Class Imbalance Problem in RSVP Tasks |
| title_short | BWGAN-GP: An EEG Data Generation Method for Class Imbalance Problem in RSVP Tasks |
| title_sort | bwgan gp an eeg data generation method for class imbalance problem in rsvp tasks |
| topic | Rapid serial visual presentation (RSVP) Wasserstein generative adversarial network (WGAN) data augmentation class imbalance problem auto-encoder |
| url | https://ieeexplore.ieee.org/document/9690467/ |
| work_keys_str_mv | AT mengxu bwgangpaneegdatagenerationmethodforclassimbalanceprobleminrsvptasks AT yuanfangchen bwgangpaneegdatagenerationmethodforclassimbalanceprobleminrsvptasks AT yijunwang bwgangpaneegdatagenerationmethodforclassimbalanceprobleminrsvptasks AT danwang bwgangpaneegdatagenerationmethodforclassimbalanceprobleminrsvptasks AT zehualiu bwgangpaneegdatagenerationmethodforclassimbalanceprobleminrsvptasks AT lijianzhang bwgangpaneegdatagenerationmethodforclassimbalanceprobleminrsvptasks |