Abstract
Recent successes of deep learning methods in various applications have inspired BCI researchers for their use in EEG classification. However, data insufficiency and high intra-and inter-subject variabilities hinder from taking their advantage of discovering complex patterns inherent in data, which can be potentially useful to enhance EEG classification accuracy. In this paper, we devise a novel framework of training a deep network by adapting samples of other subjects as a means of domain adaptation. Assuming that there are EEG trials of motor-imagery tasks from multiple subjects available, we first select a subject whose EEG signal characteristics are similar to the target subject based on their power spectral density in resting-state EEG signals. We then use EEG signals of both the selected subject (called a source subject) and the target subject jointly in training a deep network. Rather than training a single path network, we adopt a multi-path network architecture, where the shared bottom layers are used to discover common features for both source and target subjects, while the upper layers branch out into (1) source-Target subject identification, (2) label prediction optimized for a source subject, and (3) label prediction optimized for a target subject. Based on our experimental results over the BCI Competition IV-IIa dataset, we validated the effectiveness of the proposed framework in various aspects.
| Original language | English |
|---|---|
| Title of host publication | 7th International Winter Conference on Brain-Computer Interface, BCI 2019 |
| Publisher | Institute of Electrical and Electronics Engineers Inc. |
| ISBN (Electronic) | 9781538681169 |
| DOIs | |
| Publication status | Published - 2019 Feb 1 |
| Event | 7th International Winter Conference on Brain-Computer Interface, BCI 2019 - Gangwon, Korea, Republic of Duration: 2019 Feb 18 → 2019 Feb 20 |
Publication series
| Name | 7th International Winter Conference on Brain-Computer Interface, BCI 2019 |
|---|
Conference
| Conference | 7th International Winter Conference on Brain-Computer Interface, BCI 2019 |
|---|---|
| Country | Korea, Republic of |
| City | Gangwon |
| Period | 19/2/18 → 19/2/20 |
Fingerprint
Keywords
- Brain-Computer Interface
- Deep Learning
- Domain Adaptation
- Electroencephalogram (EEG)
- Motor Imagery
- Transfer Learning
ASJC Scopus subject areas
- Artificial Intelligence
- Human-Computer Interaction
- Signal Processing
- Neuroscience (miscellaneous)
Cite this
Domain Adaptation with Source Selection for Motor-Imagery based BCI. / Jeon, Eunjin; Ko, Wonjun; Suk, Heung-Il.
7th International Winter Conference on Brain-Computer Interface, BCI 2019. Institute of Electrical and Electronics Engineers Inc., 2019. 8737340 (7th International Winter Conference on Brain-Computer Interface, BCI 2019).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Domain Adaptation with Source Selection for Motor-Imagery based BCI
AU - Jeon, Eunjin
AU - Ko, Wonjun
AU - Suk, Heung-Il
PY - 2019/2/1
Y1 - 2019/2/1
N2 - Recent successes of deep learning methods in various applications have inspired BCI researchers for their use in EEG classification. However, data insufficiency and high intra-and inter-subject variabilities hinder from taking their advantage of discovering complex patterns inherent in data, which can be potentially useful to enhance EEG classification accuracy. In this paper, we devise a novel framework of training a deep network by adapting samples of other subjects as a means of domain adaptation. Assuming that there are EEG trials of motor-imagery tasks from multiple subjects available, we first select a subject whose EEG signal characteristics are similar to the target subject based on their power spectral density in resting-state EEG signals. We then use EEG signals of both the selected subject (called a source subject) and the target subject jointly in training a deep network. Rather than training a single path network, we adopt a multi-path network architecture, where the shared bottom layers are used to discover common features for both source and target subjects, while the upper layers branch out into (1) source-Target subject identification, (2) label prediction optimized for a source subject, and (3) label prediction optimized for a target subject. Based on our experimental results over the BCI Competition IV-IIa dataset, we validated the effectiveness of the proposed framework in various aspects.
AB - Recent successes of deep learning methods in various applications have inspired BCI researchers for their use in EEG classification. However, data insufficiency and high intra-and inter-subject variabilities hinder from taking their advantage of discovering complex patterns inherent in data, which can be potentially useful to enhance EEG classification accuracy. In this paper, we devise a novel framework of training a deep network by adapting samples of other subjects as a means of domain adaptation. Assuming that there are EEG trials of motor-imagery tasks from multiple subjects available, we first select a subject whose EEG signal characteristics are similar to the target subject based on their power spectral density in resting-state EEG signals. We then use EEG signals of both the selected subject (called a source subject) and the target subject jointly in training a deep network. Rather than training a single path network, we adopt a multi-path network architecture, where the shared bottom layers are used to discover common features for both source and target subjects, while the upper layers branch out into (1) source-Target subject identification, (2) label prediction optimized for a source subject, and (3) label prediction optimized for a target subject. Based on our experimental results over the BCI Competition IV-IIa dataset, we validated the effectiveness of the proposed framework in various aspects.
KW - Brain-Computer Interface
KW - Deep Learning
KW - Domain Adaptation
KW - Electroencephalogram (EEG)
KW - Motor Imagery
KW - Transfer Learning
UR - http://www.scopus.com/inward/record.url?scp=85068310816&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85068310816&partnerID=8YFLogxK
U2 - 10.1109/IWW-BCI.2019.8737340
DO - 10.1109/IWW-BCI.2019.8737340
M3 - Conference contribution
AN - SCOPUS:85068310816
T3 - 7th International Winter Conference on Brain-Computer Interface, BCI 2019
BT - 7th International Winter Conference on Brain-Computer Interface, BCI 2019
PB - Institute of Electrical and Electronics Engineers Inc.
ER -