Application of multi-task transfer learning: The combination of EA and optimized subband regularized CSP to classification of 8-channel EEG signals with small dataset.

Introduction: The volume conduction effect and high dimensional characteristics triggered by the excessive number of channels of EEG cap-acquired signals in BCI systems can increase the difficulty of classifying EEG signals and the lead time of signal acquisition. We aim to combine transfer learning to decode EEG signals in the few-channel case, improve the classification performance of the motor imagery BCI system across subject cases, reduce the cost of signal acquisition performed by the BCI system, and improve the usefulness of the system. Methods: Dataset2a from BCI CompetitionIV(2008) was used as Dataset1, and our team's self-collected dataset was used as Dataset2. Dataset1 acquired EEG signals from 9 subjects using a 22-channel device with a sampling frequency of 250 Hz. Dataset2 acquired EEG signals from 10 healthy subjects (8 males and 2 females; age distribution between 21-30 years old; mean age 25 years old) using an 8-channel system with a sampling frequency of 1000 Hz. We introduced EA in the data preprocessing process to reduce the signal differences between subjects and proposed VFB-RCSP in combination with RCSP and FBCSP to optimize the effect of feature extraction. Results: Experiments were conducted on Dataset1 with EEG data containing only 8 channels and achieved an accuracy of 78.01 and a kappa coefficient of 0.54. The accuracy exceeded most of the other methods proposed in recent years, even though the number of channels used was significantly reduced. On Dataset 2, an accuracy of 59.77 and a Kappa coefficient of 0.34 were achieved, which is a significant improvement compared to other poorly improved classical protocols. Discussion: Our work effectively improves the classification of few-channel EEG data. It overcomes the dependence of existing algorithms on the number of channels, the number of samples, and the frequency band, which is significant for reducing the complexity of BCI models and improving the user-friendliness of BCI systems.

Supervised and Semisupervised Manifold Embedded Knowledge Transfer in Motor Imagery-Based BCI.

A long calibration procedure limits the use in practice for a motor imagery (MI)-based brain-computer interface (BCI) system. To tackle this problem, we consider supervised and semisupervised transfer learning. However, it is a challenge for them to cope with high intersession/subject variability in the MI electroencephalographic (EEG) signals. Based on the framework of unsupervised manifold embedded knowledge transfer (MEKT), we propose a supervised MEKT algorithm (sMEKT) and a semisupervised MEKT algorithm (ssMEKT), respectively. sMEKT only has limited labelled samples from a target subject and abundant labelled samples from multiple source subjects. Compared to sMEKT, ssMEKT adds comparably more unlabelled samples from the target subject. After performing Riemannian alignment (RA) and tangent space mapping (TSM), both sMEKT and ssMEKT execute domain adaptation to shorten the differences among subjects. During domain adaptation, to make use of the available samples, two algorithms preserve the source domain discriminability, and ssMEKT preserves the geometric structure embedded in the labelled and unlabelled target domains. Moreover, to obtain a subject-specific classifier, sMEKT minimizes the joint probability distribution shift between the labelled target and source domains, whereas ssMEKT performs the joint probability distribution shift minimization between the unlabelled target domain and all labelled domains. Experimental results on two publicly available MI datasets demonstrate that our algorithms outperform the six competing algorithms, where the sizes of labelled and unlabelled target domains are variable. Especially for the target subjects with 10 labelled samples and 270/190 unlabelled samples, ssMEKT shows 5.27% and 2.69% increase in average accuracy on the two abovementioned datasets compared to the previous best semisupervised transfer learning algorithm (RA-regularized common spatial patterns-weighted adaptation regularization, RA-RCSP-wAR), respectively. Therefore, our algorithms can effectively reduce the need of labelled samples for the target subject, which is of importance for the MI-based BCI application.

Use of phase-locking value in sensorimotor rhythm-based brain-computer interface: zero-phase coupling and effects of spatial filters.

Phase-locking value (PLV) is a potentially useful feature in sensorimotor rhythm-based brain-computer interface (BCI). However, volume conduction may cause spurious zero-phase coupling between two EEG signals and it is not clear whether PLV effects are independent of spectral amplitude. Volume conduction might be reduced by spatial filtering, but it is uncertain what impact this might have on PLV. Therefore, the goal of this study was to explore whether zero-phase PLV is meaningful and how it is affected by spatial filtering. Both amplitude and PLV feature were extracted in the frequency band of 10-15 Hz by classical methods using archival EEG data of 18 subjects trained on a two-target BCI task. The results show that with right ear-referenced data, there is meaningful long-range zero-phase synchronization likely involving the primary motor area and the supplementary motor area that cannot be explained by volume conduction. Another novel finding is that the large Laplacian spatial filter enhances the amplitude feature but eliminates most of the phase information seen in ear-referenced data. A bipolar channel using phase-coupled areas also includes both phase and amplitude information and has a significant practical advantage since fewer channels required.

EEG based zero-phase phase-locking value (PLV) and effects of spatial filtering during actual movement.

Phase-locking value (PLV) is a well-known feature in sensorimotor rhythm (SMR) based BCI. Zero-phase PLV has not been explored because it is generally regarded as the result of volume conduction. Because spatial filters are often used to enhance the amplitude (square root of band power (BP)) feature and attenuate volume conduction, they are frequently applied as pre-processing methods when computing PLV. However, the effects of spatial filtering on PLV are ambiguous. Therefore, this article aims to explore whether zero-phase PLV is meaningful and how this is influenced by spatial filtering. Based on archival EEG data of left and right hand movement tasks for 32 subjects, we compared BP and PLV feature using data with and without pre-processing by a large Laplacian. Results showed that using ear-referenced data, zero-phase PLV provided unique information independent of BP for task prediction which was not explained by volume conduction and was significantly decreased when a large Laplacian was applied. In other words, the large Laplacian eliminated the useful information in zero-phase PLV for task prediction suggesting that it contains effects of both amplitude and phase. Therefore, zero-phase PLV may have functional significance beyond volume conduction. The interpretation of spatial filtering may be complicated by effects of phase.