Transfer Learning for Brain–Computer Interfaces: A Euclidean Space Data Alignment Approach
Open accessThe DOI matches its Crossref record; 50 of the 52 references it cites were found in a registry or an index; 9 of the 30 sources its claims cite were checked for support.
Abstract
OBJECTIVE: This paper targets a major challenge in developing practical electroencephalogram (EEG)-based brain-computer interfaces (BCIs): how to cope with individual differences so that better learning performance can be obtained for a new subject, with minimum or even no subject-specific data? METHODS: We propose a novel approach to align EEG trials from different subjects in the Euclidean space to make them more similar, and hence improve the learning performance for a new subject. Our approach has three desirable properties: first, it aligns the EEG trials directly in the Euclidean space, and any signal processing, feature extraction, and machine learning algorithms can then be applied to the aligned trials; second, its computational cost is very low; and third, it is unsupervised and does not need any label information from the new subject. RESULTS: Both offline and simulated online experiments on motor imagery classification and event-related potential classification verified that our proposed approach outperformed a state-of-the-art Riemannian space data alignment approach, and several approaches without data alignment. CONCLUSION: The proposed Euclidean space EEG data alignment approach can greatly facilitate transfer learning in BCIs. SIGNIFICANCE: Our proposed approach is effective, efficient, and easy to implement. It could be an essential pre-processing step for EEG-based BCIs.
Overview
CoChat Summary
TL;DR
The paper proposes Euclidean-space alignment (EA), an unsupervised transfer-learning preprocessing step that aligns EEG trials from different subjects in Euclidean space to reduce individual differences for BCI classification. Across offline and simulated online experiments on motor imagery and ERP datasets, EA generally outperformed Riemannian-space alignment and no-alignment baselines while being much faster.
Key findings
- In offline unsupervised MI classification, EA-CSP-LDA achieved average accuracies of 79.79% on MI Dataset 1 and 73.53% on MI Dataset 2a, versus RA-MDRM at 68.07% and 70.91%, CSP-LDA at 59.71% and 67.75%, and MDRM at 54.36% and 57.18%.
- In offline ERP classification, EA-xDAWN-SVM achieved an average balanced classification accuracy (BCA) of 68.80%, compared with EA-SVM 67.85%, SVM 64.64%, and xDAWN-SVM 64.60%; EA-SVM significantly outperformed SVM (p=0.0386) and EA-xDAWN-SVM significantly outperformed xDAWN-SVM (p=0.0449).
- In simulated online MI classification, EA-CSP-LDA outperformed RA-MDRM on 12 out of 16 subjects; in simulated online ERP classification, EA-xDAWN-SVM outperformed RA-MDRM on all 11 subjects.
- EA-CSP-LDA was 3.6–19.5 times faster than RA-MDRM; for example, on MI Dataset 1 it took 0.3864 s versus 7.5326 s, and on MI Dataset 2a it took 0.2405 s versus 0.8766 s.
Methods
- The proposed EA aligns time-domain EEG trials in Euclidean space by using the arithmetic mean of all covariance matrices from a subject as the reference matrix, then transforming trials so that the mean covariance matrix of each subject becomes the identity matrix; it requires no labels from the new subject.
- Experiments used two MI datasets from BCI Competition IV (Dataset 1: 7 subjects, 59-channel EEG, 100 trials per class in calibration; Dataset 2a: 9 subjects, 22-channel EEG plus 3-channel EOG, 72 trials per class for left- and right-hand MI) and one ERP RSVP dataset from PhysioNet (11 subjects, 8 channels, 5 Hz Session a, target-to-non-target ratio around 1:9).
- Evaluation included offline unsupervised leave-one-subject-out cross-validation and simulated online supervised classification, with CSP-LDA, MDRM, SVM, and xDAWN-SVM pipelines; paired-sample t-tests and area under the curve (AUC) were used for statistical comparisons.
Limitations
- The paper states that EA only compensates for covariate shift and ignores prior probability shift and concept shift; it may even increase concept shift, and bad trials or outliers can corrupt the reference matrix. It also states that the simulated online experiments are not identical to real online experiments.
- The text makes evident that EA degraded accuracy for some subjects (e.g., Subjects 4 and 9 in MI Dataset 2a, and some ERP subjects), and that RA-MDRM could not be applied to offline unsupervised ERP classification because it requires labeled non-target and target trials from the new subject.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
4 Valid · 8 Not valid · 15 Not checked · 89 Own results and interpretations
Every source the paper cites for this claim was checked and supports it.
In the paper
“So Barachant and Congedo [2] proposed a novel approach to augment the ERP trials so that their covariance matrices can also encode the temporal information.”
What the sources say
- Supportedcites [2] [2] A. Barachant and M. Congedo, 'A plug & play P300 BCI using information geometry,' arXiv: 1409.0107 , 2014. - source
Section B (pp. 3–4) describes a prototyped ERP response obtained by averaging single-trial responses, then concatenating it with each trial to form a super trial. The resulting covariance matrices include a cross-covariance block that carries information about the ERP waveform’s temporal structure. (LLM, conf=0.94)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
Every source the paper cites for this claim was checked and supports it.
In the paper
“However, the main discriminative information of ERP trials is carried temporally rather than spatially.”
What the sources say
- Supportedcites [2] [2] A. Barachant and M. Congedo, 'A plug & play P300 BCI using information geometry,' arXiv: 1409.0107 , 2014. - source
Section B (p. 3) says ERP discriminative information lies in the potential’s waveform rather than the spatial distribution of its variance. It also says ignoring temporal structure loses information needed for classification, supporting the paraphrase. (LLM, conf=0.93)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our proposed EA-CSP-LDA was 3.6-19.5 times faster than RA-MDRM, and also it had much smaller standard deviation.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“On average RI-MDRM outperformed RR-MDRM, and EI-CSP-LDA outperformed ER-CSP-LDA, on both datasets, suggesting that estimating the reference matrix from all imagery trials would be better than using all resting trials.”
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
A result the paper reports from its own data — it cites no other source for it, so there is none to check. The quote shows it is in the paper.
In the paper
“Our proposed EA-CSP-LDA was 3.6-19.5 times faster than RA-MDRM, and also it had much smaller standard deviation.”
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
47 more claims
Create a free account to see all 116 claims and to verify the 4 references still open.
- It may state two separate claims as one. 47
- Probably not a substantial claim. 30
- The quote only partly states this claim. 1
Citations
Cites
LIBSVM: A library for support vector machines
Chang, C.-C., & Lin, C.-J. (2011) · ACM Transactions on Intelligent Systems and Technology · doi:10.1145/1961189.1961199 · Cited by 40,918 · cited 1×
“LibSVM [6] with a linear kernel was used for classification.”
Method(2008) · Journal of Machine Learning Research · Cited by 33,051
Delorme, A., & Makeig, S. (2004) · Journal of Neuroscience Methods · doi:10.1016/j.jneumeth.2003.10.009 · Cited by 24,371
Pan, S. J., & Yang, Q. (2010) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2009.191 · Cited by 23,387 · cited 2×
“This idea has been widely used in TL [24], [31], [40].”
BackgroundMethodBook Review: Pattern Recognition and Machine Learning
Nasrabadi, N. M. (2007) · Journal of Electronic Imaging · doi:10.1117/1.2819119 · Cited by 22,066
A Simple Sequentially Rejective Multiple Test Procedure
(1979) · Scandinavian Journal of Statistics · doi:10.2307/4615733 · Cited by 21,526 · cited 1×
“Hence, we performed Holm-Bonferroni corrections [12] on the three p-values of each dataset.”
MethodGoldberger, A. L., Amaral, L. A. N., Glass, L., Hausdorff, J. M., Ivanov, P. C., Mark, R. G., Mietus, J. E., Moody, G. B., Peng, C.-K., & Stanley, H. E. (2000) · Circulation · doi:10.1161/01.cir.101.23.e215 · Cited by 14,809
Pattern Recognition and Machine Learning
(2006) · doi:10.1007/978-0-387-45528-0 · Cited by 9,824 · cited 1×
“Usually the EEG signals are first band-pass filtered and spatially filtered to increase the signal-to-noise ratio, and then discriminative features are extracted, which are next fed into machine learning algorithms such as Linear Discriminant Analysis (LDA) and Support Vector Machine (SVM) [3] for classification.”
MethodBrain–computer interfaces for communication and control
Wolpaw, J. R., Birbaumer, N., McFarland, D. J., Pfurtscheller, G., & Vaughan, T. M. (2002) · Clinical Neurophysiology · doi:10.1016/s1388-2457(02)00057-3 · Cited by 7,927
On the Kolmogorov-Smirnov Test for Normality with Mean and Variance Unknown
Lilliefors, H. W. (1967) · Journal of the American Statistical Association · doi:10.2307/2283970 · Cited by 4,530
Cited by
Items 201–300 of 520
Wang, H., Han, H., Gan, J. Q., & Wang, H. (2024) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2024.3463737 · Cited by 11
“EA [24] assumes that the source and target domains should have consistent average covariance matrices, representing the distribution between each channel’s time series.”
InfluentialZuo, M., Chen, X., & Sui, L. (2025) · Brain Sciences · doi:10.3390/brainsci15010075 · Cited by 10
Latent alignment in deep learning models for EEG decoding
Bakas, S., Ludwig, S., Adamos, D. A., Laskaris, N., Panagakis, Y., & Zafeiriou, S. P. (2025) · Journal of Neural Engineering · doi:10.1088/1741-2552/adb336 · Cited by 25
“These are most often applied on the input signal [4–6], but the personalized standardization of classification features has also been performed [7, 8].”
Zhu, J. L., Xu, G., Lin, Z., Long, J., Zhou, T., Sheng, B., & Yang, X. (2025) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2025.3604283 · Cited by 12
“For example, although EA [40] has the lowest computational overhead, its accuracy is only slightly higher than EEGNet [43].”
InfluentialA multi-scale dual-stream fusion network for high-accuracy sEMG-based gesture classification
He, D., Liu, W., Yan, H., Yun, Z., & Jiang, B. (2026) · Scientific Reports · doi:10.1038/s41598-025-34909-0 · Cited by 14
“Studies have shown that different hand gestures are often associated with distinct frequency patterns in sEMG signals, enhancing the accuracy of classification models 37,56 .”
Feng, J., Li, Y., Jiang, C., Liu, Y., Li, M., & Hu, Q. (2022) · Frontiers in Human Neuroscience · doi:10.3389/fnhum.2022.1068165 · Cited by 9
“The EA-CSP-LDA algorithm reduces the difference of cross-subject signals based on the Euclidean space data alignment approach.”
MethodInfluentialTong, J., Wei, X., Dong, E., Sun, Z., Du, S., & Duan, F. (2022) · Journal of Neural Engineering · doi:10.1088/1741-2552/ac9a01 · Cited by 8
“…mean value) for the training data and the updating reference matrix points for the test data, we use the data shifting method proposed by He and Wu [53] to transform the training data and test data distribution as shown in equation (18): where C R can be the reference matrix point of training data…”
Segment alignment based cross-subject motor imagery classification under fading data
Wan, Z., Yang, R., Huang, M., Alsaadi, F. E., Sheikh, M. M., & Wang, Z. (2022) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2022.106267 · Cited by 8
Adversarial Attacks and Defenses in Physiological Computing: A Systematic ReviewPreprint
Wu, D., Xu, J., Fang, W., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2102.02729 · Cited by 8
Self-citationGuo, S., Wang, Y., Liu, Y., Zhang, X., & Tang, B. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.107756 · Cited by 7
Wu, H., & Wu, D. (2022) · Brain Science Advances · doi:10.26599/bsa.2022.9050001 · Cited by 7
“In this section, we first introduce preprocessing approaches, such as filter and Euclidean space alignment (EA) [14].”
MethodSelf-citationFacilitating Applications of SSVEP-Based BCIs by Within-Subject Information Transfer
Liu, X., Liu, B., Dong, G., Gao, X., & Wang, Y. (2022) · Frontiers in Neuroscience · doi:10.3389/fnins.2022.863359 · Cited by 10
“For instance, He and Wu (He and Wu, 2020) applied the subspace alignment to the motor imagery paradigm by aligning the EEG of different subjects in the Euclidean space and then applying it to new subjects.”
BackgroundLi, J., Zhou, Y., Liu, T., Jung, T., Wan, X., Duan, D., Li, D., Yu, H., Song, H., Dong, X., et al. (2024) · Expert Systems with Applications · doi:10.1016/j.eswa.2024.125089 · Cited by 7
A novel multi-morphological representation approach for multi-source EEG signals
Gao, Y., Liu, Y., Meng, M., Fang, F. Z., Houston, M. J., & Zhang, Y. (2024) · Neurocomputing · doi:10.1016/j.neucom.2024.129010 · Cited by 10
A novel contrastive Dual-Branch Network (CDB-Net) for robust EEG-Based Alzheimer’s disease diagnosis
Alouani, Z., Gannour, O. E., Saleh, S., et al. (2025) · Brain Research · doi:10.1016/j.brainres.2025.149863 · Cited by 7
Milyani, A. H., & Attar, E. T. (2025) · Frontiers in Human Neuroscience · doi:10.3389/fnhum.2025.1668935 · Cited by 12
Zhang, J., Fang, J., Liu, S., Liu, D., Wu, H., & Long, J. (2024) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2024.3451593 · Cited by 13
“2) Comparing Methods: Eight well-established EEG classification methods are selected for method comparison, which are EEGNet [18], 1D-CNN [38], EEG-DG [65], SACNN [43], TS-TCC [66], CSP-LDA [23], EA-CSP-LDA [67], CA [4], CA-JGSA [68], CA-JDA [69], and CA-CORAL [70].”
Yang, S., Huang, Z., & Luo, T.-J. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.127312 · Cited by 8
EEG multi-domain feature transfer based on sparse regularized Tucker decomposition
Gao, Y., Zhang, C., Huang, J., & Meng, M. (2023) · Cognitive Neurodynamics · doi:10.1007/s11571-023-09936-0 · Cited by 8
Long, T., Wan, M., Jian, W., Dai, H., Nie, W., & Xu, J. (2023) · Frontiers in Human Neuroscience · doi:10.3389/fnhum.2023.1143027 · Cited by 7
“The EEG signal was acquired by wet electrode method, and the acquisition device was NeuSen W wireless digital EEG acquisition system produced by Neuracle Technology (Changzhou) Co., Ltd. with 8 electrode channels (FC3, FCz, FC4, C3, Cz, C4, CP3, and CP4), with CPz as the reference electrode and AFz as the ground elect…”
MethodInfluentialChannel and Trials Selection for Reducing Covariate Shift in EEG-based Brain-Computer Interfaces
He, H., & Wu, D. (2019) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc.2019.8914414 · Cited by 7
“Our recent work [24] discussed different choices of the reference matrix, and demonstrated that the Riemannian mean of the imagery trials outperformed that of the resting trials.”
MethodSelf-citationAI and Computer Vision Technologies for Metaverse
The, T. H., Pham, Q., Pham, X., Do‐Duy, T., & Gadekallu, T. R. (2023) · Wiley eBooks · doi:10.1002/9781394160013.ch5 · Cited by 8
Chen, P., Liu, X., Ma, C., Wang, H., Yang, X., Grebogi, C., Gu, X., & Gao, Z. (2025) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2025.3525577 · Cited by 16
“To reduce cross-domain variability, many transfer learning methods have been proposed for EEG-based BCI systems [16], [32], [33].”
InfluentialEEGNet-based multi-source domain filter for BCI transfer learning
Li, M., Li, J., Song, Z., Deng, H., Xu, J., Xu, G., & Liao, W. (2023) · Medical & Biological Engineering & Computing · doi:10.1007/s11517-023-02967-z · Cited by 7
“He and Wu proposed a feature-based DA model that aligned the target subject with source subjects’ features in Euclidean space for training, which extracted the domain-invariant features well [28].”
D-FaST: Cognitive Signal Decoding With Disentangled Frequency–Spatial–Temporal Attention
Chen, W., Wang, C., Xu, K., Yuan, Y., Bai, Y., & Zhang, D. (2024) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2024.3370261 · Cited by 8
“Subsequent steps included band-pass filtering in the frequency range of 4 to 38Hz, followed by a normalization [60] and European alignment [61].”
MethodKilani, S., Aghili, S. N., Fathi, Y., & Sburlea, A. I. (2024) · Frontiers in Neuroscience · doi:10.3389/fnins.2024.1360709 · Cited by 6
Group-member selection for RSVP-based collaborative brain-computer interfaces
Si, Y., Wang, Z., Xu, G., Wang, Z., Xu, T., Zhou, T., & Hu, H. (2024) · Frontiers in Neuroscience · doi:10.3389/fnins.2024.1402154 · Cited by 6
“The Euclidean space data alignment (EA) (He and Wu, 2020) was further employed to address the issue of inter-trial variability in cross-session conditions.”
BackgroundMethodXiao, F., Dai, S., & Guo, Z. (2024) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2024.107132 · Cited by 5
Bispectrum-based hybrid neural network for motor imagery classification
Liu, C., Jin, J., Daly, I. P., Sun, H., Huang, Y., Wang, X., & Cichocki, A. S. (2022) · Journal of Neuroscience Methods · doi:10.1016/j.jneumeth.2022.109593 · Cited by 5
LDER: a classification framework based on ERP enhancement in RSVP task
Cui, Y., Xie, S., Xie, X., Zheng, D., Tang, H., Duan, K., Chen, X., & Jiang, Y. (2023) · Journal of Neural Engineering · doi:10.1088/1741-2552/acd95d · Cited by 7
“He and Wu developed a Euclidean space EEG signal alignment method that can make the distributions of different data more similar [23].”
MethodZhu, F., Cai, J., Zheng, H., Liang, Z., & Zhang, Y. (2024) · Review of Scientific Instruments · doi:10.1063/5.0208524 · Cited by 7
Motor Imagery Classification Based on Subject to Subject Transfer in Riemannian Manifold
Singh, A. S. G., Lal, S. P., & Guesgen, H. W. (2019) · International Winter Conference on Brain-Computer Interface (BCI) · doi:10.1109/iww-bci.2019.8737256 · Cited by 6
“Riemannian distance follows congruence invariance [13], which means the distance measure remains unchanged under linear invertible transformation.”
BackgroundAn efficient domain adaptation framework for cross-subject EEG classification
Luo, T.-J. (2023) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm58861.2023.10385973 · Cited by 8
“Researchers treated the samples as multivariate time-series, and perform alignment on the covariance matrices, since the covariance represents the samples distribution, such as Riemannian alignment (RA) [6] and Euclidean alignment (EA) [7] method.”
MethodKim, K., Kim, K., & Lee, S.-B. (2024) · Neurocomputing · doi:10.1016/j.neucom.2024.127805 · Cited by 5
MI-DGHCL: Motor imagery EEG domain generalization via hyperbolic contrastive learning
Chen, J., Pi, D., Gao, F., et al. (2026) · Expert Systems with Applications · doi:10.1016/j.eswa.2026.131477 · Cited by 5
Guo, Z., Feng, Y., Ma, M., Sun, Y., & Xia, L. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.107706 · Cited by 5
Neurological Disorder Diagnosis and Treatment Using BCIs
Nithya, V., & Leena, B. (2025) · IGI Global eBooks · doi:10.4018/979-8-3693-9445-8.ch020 · Cited by 5
MVCNet: Multi-view contrastive network for motor imagery classification
Wang, Z., Li, S., Chen, X., & Wu, D. (2025) · Knowledge-Based Systems · doi:10.1016/j.knosys.2025.114205 · Cited by 17
“Euclidean Alignment (EA) [47], an effective unsupervised EEG data alignment approach [6], was utilized after pre-processing.”
Self-citationGlobal research trends in reverse vaccinology from 2000 to 2021: A bibliometric analysis
Salod, Z., & Mahomed, O. H. (2023) · Informatics in Medicine Unlocked · doi:10.1016/j.imu.2023.101313 · Cited by 7
Transfer Learning for P300 Brain-Computer Interfaces by Joint Alignment of Feature Vectors
Altindış, F., Banerjee, A., Phlypo, R., Yılmaz, B., & Congedo, M. (2023) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2023.3299837 · Cited by 9
“In [21] the authors whitened the arithmetic average of the covariance matrices estimated on the EEG trials of each subject.”
BackgroundResultYılmaz, Y. (2023) · Neural Computing and Applications · doi:10.1007/s00521-023-08995-y · Cited by 6
“Learning takes place in biological systems by modifying the synaptic links connecting neurons [12].”
BackgroundMethodMotor Imagery Decoding Using Ensemble Curriculum Learning and Collaborative Training
Zoumpourlis, G., & Patras, I. (2024) · International Winter Conference on Brain-Computer Interface (BCI) · doi:10.1109/bci60775.2024.10480476 · Cited by 16
“Other methods that have been occasionaly used for multi-subject EEG training are Euclidean/Riemannian Alignment (EA/RA) [12], [17], [33], which are powerful baselines for learning domain-invariant representations.”
MethodSignal alignment for cross-datasets in P300 brain-computer interfaces
Song, M., Gwon, D., Jun, S. C., & Ahn, M. (2024) · Journal of Neural Engineering · doi:10.1088/1741-2552/ad430d · Cited by 7
“Also, various types of means can be used, such as the Euclidean mean for Euclidean alignment [42], the Riemannian mean for Riemannian alignment [43], and the log-Euclidean mean for log-Euclidean alignment [41].”
MethodWeight-freezing: A motor imagery inspired regularization approach for EEG classification
Miao, Z., & Zhao, M. (2024) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2024.107015 · Cited by 11
Ma, S., & Zhang, D. (2024) · Sensors · doi:10.3390/s24217080 · Cited by 7
Chen, J., Yi, W., & Wang, D. (2021) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc46164.2021.9629728 · Cited by 7
“The Euclidean alignment method is used to preprocess the MI-EEG signal to improve the decoding performance by transforming the mean covariance matrix of MI-EEG signals to an identity matrix [16].”
MethodCalibration-Free Transfer Learning for EEG-Based Cross-Subject Motor Imagery Classification
Wang, Y., Wang, J., Wang, W., Su, J., & Hou, Z. (2023) · IEEE International Conference on Automation Science and Engineering (CASE) · doi:10.1109/case56687.2023.10260440 · Cited by 7
Lin, W., Chen, W.-J., Chen, Y., Liang, H.-Y., Lu, C., & Lin, Y. (2023) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2023.3323902 · Cited by 6
“Second, recent BCI studies have reported the success of advanced spectral feature mining [44], deep learning [45], and transfer learning [46], facilitating the modeling of EEG activity in response to tasks of interest.”
BackgroundDiscriminative Joint Probability Maximum Mean Discrepancy (DJP-MMD) for Domain AdaptationPreprint
Zhang, W., & Wu, D. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1912.00320 · Cited by 5
“Domain adaptation (DA), or transfer learning, has shown promising performance in handling these challenges [2]–[8], by transferring knowledge from a labeled source domain to a new unlabeled or partially labeled target domain.”
Self-citationA Novel Prediction Model for Multimodal Medical Data Based on Graph Neural Networks
Zhang, L., Li, T., Cui, H., Zhang, Q., Jiang, Z., Li, J., Welsch, R. E., & Jia, Z. (2025) · Machine Learning and Knowledge Extraction · doi:10.3390/make7030092 · Cited by 4
“In Euclidean space data, for example, text and speech data are one-dimensional sequence data structures [20], and image data [21,22] are typically rectangular structures.”
Enhanced BCI Performance using Diffusion Model for EEG Generation
Zhong, Y., Yao, L., & Wang, Y. (2024) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc53108.2024.10782900 · Cited by 5
“To mitigate variations arising from diverse subjects and recording devices, signal standardization was implemented for each trial [11]: where M represents the Euclidean mean derived from the covariance matrices of all trials of a specific subject.”
Liang, Z., Zheng, Z., Chen, W., Ma, X., Pei, Z., & Sun, X. (2025) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2025.110340 · Cited by 6
An, S., Kang, M., Kim, S., Chikontwe, P., Shen, L., & Park, S. H. (2025) · Information Fusion · doi:10.1016/j.inffus.2025.103501 · Cited by 5
Motor imagery decoding network with multisubject dynamic transfer
Li, Z., Li, M., & Yang, Y. (2025) · Brain Informatics · doi:10.1186/s40708-025-00267-w · Cited by 5
“…methods were compared, including traditional methods such as FBCSP [32], traditional transfer learning methods RA-MDRM [33] and EA-CSP-LDA [34], deep learning (DL) methods EEGNet [14] and ConvNet [13], the supervised domain adaptation method MI-DABAN [25], and unsupervised domain…”
Lian, X., Liu, C., Gao, C., Deng, Z., Guan, W., & Gong, Y. (2025) · Brain Sciences · doi:10.3390/brainsci15080877 · Cited by 4
“Additionally, data alignment techniques, such as Euclidean alignment [46], could be employed prior to model training to reduce inter-subject distributional discrepancies.”
Enhancing multi-paradigm EEG signal classification in cross-subject settings using optimal transport
Morvay, B. T., Torma, S., Pitrik, J., & Szegletes, L. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.108892 · Cited by 6
Neural decoding for EEG-BCI: from conventional machine learning to deep learning models
Ding, Y., Ma, X., Zhang, P., Tang, Y., Zhao, Z., Chen, D., Wu, D., & Tang, Z. (2026) · Brain Hemorrhages · doi:10.1016/j.hest.2026.01.002 · Cited by 8
Self-citationLiu, D., Ding, Q., Tong, C., Ai, J., & Wang, F. (2024) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2024.3491096 · Cited by 6
Yang, C., Zhang, Y., Wang, S., et al. (2025) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2025.2464097 · Cited by 4
“He and Wu (2020) proposed a Euclidean Space Alignment method to adjust EEG trials from different subjects in the Euclidean space, making them more similar and thereby improving the learning performance for new subjects.”
EDANet: Efficient domain-adaptive attention neural network for EEG classification of motor imagery
Yang, J., Wang, L., Cai, W., Zhang, L., Xie, C., & Wang, Z. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.128783 · Cited by 5
Hu, H., Zhou, Z., Zhang, Z., & Yuan, W. (2025) · Electronics · doi:10.3390/electronics14142853 · Cited by 5
Multi-Source Discriminant Dynamic Domain Adaptation for Cross-Subject Motor Imagery EEG Recognition
Gong, Y., Shi, K., Niu, X., Yang, L., Yang, X., & Chen, Z. (2025) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2025.3610446 · Cited by 11
“For the BCI IV 1 and BCI IV 2a datasets, the preprocessing procedure follows the approach described in [27].”
InfluentialSingle-Trial MEG Data Can Be Denoised Through Cross-Subject Predictive Modeling
Ravishankar, S., Toneva, M., & Wehbe, L. (2021) · Frontiers in Computational Neuroscience · doi:10.3389/fncom.2021.737324 · Cited by 6
“Of the TL approaches, Euclidean-space alignment (EA) (He and Wu, 2019) is closest to our goal because it aligns the data in Euclidean space, after which further analysis can be performed.”
BackgroundImproved motor imagery training for subject’s self-modulation in EEG-based brain-computer interface
Xu, Y., Jie, L., Jian, W., Yi, W., Yin, H., & Peng, Y. (2024) · Frontiers in Human Neuroscience · doi:10.3389/fnhum.2024.1447662 · Cited by 4
“…1 to N remove dis _ do remove the j -th entry of σ i and then obtain new variance σ σ i ( Then, Riemannian alignment (RA) (Zanini et al., 2018; He and Wu, 2019; Xu et al., 2021) and common spatial pattern (CSP) method (Ramoser et al., 2000) were successively executed to extract valuable…”
MethodCai, Z., Gao, Y., Fang, F. Z., Zhang, Y., & Du, S. (2024) · Journal of Neuroscience Methods · doi:10.1016/j.jneumeth.2024.110332 · Cited by 5
Sparse knowledge sharing (SKS) for privacy-preserving domain incremental seizure detection
An, J., Peng, R., Du, Z., Liu, H., Hu, F., Shu, K., & Wu, D. (2025) · Journal of Neural Engineering · doi:10.1088/1741-2552/adb998 · Cited by 4
“To minimize domain distribution shifts and facilitate knowledge sharing among domains, we employ EA [33] to align data from different domains.”
Self-citationZhan, Q., Wang, L., Ren, L., & Huang, X. (2022) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2022.106220 · Cited by 4
Manifold Trial Selection to Reduce Negative Transfer in Motor Imagery-based Brain–Computer Interface
Liang, Z., Zheng, Z., Chen, W., Wang, J., Zhang, J., Chen, J., & Chen, Z. (2021) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros51168.2021.9636137 · Cited by 8
“On the basis of RA, [21] proposed a center alignment method in Euclidean space, called Euclidean alignment (EA).”
MethodResultInfluentialFordson, H. P., Xing, X., Guo, K., & Xu, X. (2022) · Frontiers in Neuroscience · doi:10.3389/fnins.2022.865201 · Cited by 4
“The studies of He and Wu (2020) proposed an approach to align EEG trials from several subjects in Euclidean space by reducing variations that improve the learning performance of new incoming subjects.”
BackgroundTransfer Learning for Brain-Computer Interfaces: A Complete PipelinePreprint
Wu, D., Peng, R., Huang, J., & Zeng, Z. (2020) · arXiv (Cornell University) · Cited by 6
Self-citationTransfer Learning for Motor Imagery Based Brain-Computer Interfaces: A Complete PipelinePreprint
Wu, D., Jiang, X., Peng, R., et al. (2020) · arXiv (Cornell University) · doi:10.48550/arxiv.2007.03746 · Cited by 4
“Different EEG trial alignment approaches have been proposed recently [22], [23], [25], [28], [29].”
BackgroundMethodInfluentialSelf-citation[The research status and development trends of brain-computer interfaces in medicine].
Chen, Q., Yuan, T., Zhang, L., Gong, J., Fu, L., Han, X., Ruan, M., & Yu, Z. (2023) · PubMed Central · doi:10.7507/1001-5515.202303038 · Cited by 4
Vishwanath, M., Cao, S., Dutt, N., Rahmani, A. M., Lim, M. M., & Cao, H. (2025) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2025.107948 · Cited by 3
Mohanapriya, D., Chairman, M., Manojkumar, B., et al. (2025) · IGI Global eBooks · doi:10.4018/979-8-3693-9445-8.ch015 · Cited by 3
Chen, X., Fu, Z., Zhang, P., Chen, X., & Huang, J. (2025) · IEEE Transactions on Biomedical Engineering · doi:10.1109/tbme.2025.3586870 · Cited by 5
“However, these algorithms primarily focus on initial de-coder construction, failing to effectively address the neural signal characteristic drift caused by dynamic changes in the electrode-tissue interface [14], [15].”
Jang, Y.-J., Jeong, J., Kim, Y. K., et al. (2025) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2025.3595379 · Cited by 3
“For instance, He and Wu [38] proposed the Euclidean-space alignment (EA), which improves transfer learning by aligning the covariance matrix distributions across subjects.”
Zhao, Z., Cao, Y., Yu, H., & Huang, J. (2025) · Expert Systems with Applications · doi:10.1016/j.eswa.2025.129248 · Cited by 3
Mishra, R., Agrawal, R. K., & Kirar, J. S. (2025) · Cognitive Neurodynamics · doi:10.1007/s11571-025-10337-8 · Cited by 3
“(He and Wu 2019) proposed Euclidean alignment (EA), which converts the average covariance matrix of all subjects into a unit matrix and improves clas - sification accuracy by minimizing subject differences.”
Personalized real-time inference of momentary excitability from human EEG
Haxel, L., Ahola, O., Kapoor, J., Ziemann, U., & Macke, J. H. (2025) · NeuroImage · doi:10.1016/j.neuroimage.2025.121547 · Cited by 4
Cross-subject motor imagery EEG signal classification based on meta-transfer learning
Li, H., Liu, J., & Li, J. (2026) · Computer Methods in Biomechanics & Biomedical Engineering · doi:10.1080/10255842.2026.2626477 · Cited by 5
Jiang, X., Yang, L., Xu, Y., Song, Z., Wang, Y., & He, Y. (2026) · Alexandria Engineering Journal · doi:10.1016/j.aej.2026.04.006 · Cited by 3
Parameter transfer and Riemannian space coordinate alignment for EEG intention recognition
Li, X., Tang, X., Liu, J., Liu, Q., Wang, H., & Liu, K. (2024) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2024.106044 · Cited by 3
Yang, C., Wang, S., Peng, Y., Zhang, Z., Kong, L., Zhou, C., Tao, Y., & Chen, X. (2024) · Journal of Engineering Design · doi:10.1080/09544828.2024.2326396 · Cited by 3
“He et al. (He and Wu 2020) proposed a Euclidean Space data alignment approach to align EEG trials from different subjects in the Euclidean space to make them more similar, and hence improve the learning performance for a new subject.”
BackgroundWeighted common spatial pattern based adaptation regularization for multi-source EEG time series
Han, R., Li, Z., Zhang, Y., Meng, X., Wang, Z., & Dong, H. (2024) · Computers & Electrical Engineering · doi:10.1016/j.compeleceng.2024.109680 · Cited by 3
Emotion Analysis AI Model for Sensing Architecture Using EEG
Ji, S., Kim, M. K., & Jun, H. J. (2025) · Applied Sciences · doi:10.3390/app15052742 · Cited by 6
Data alignment based adversarial defense benchmark for EEG-based BCIs
Chen, X., Jia, T., & Wu, D. (2025) · Neural Networks · doi:10.1016/j.neunet.2025.107516 · Cited by 12
Self-citationDynamic and low-dimensional modeling of brain functional connectivity on Riemannian manifolds
Wang, M., Wang, Y., & Yang, Y. (2025) · NeuroImage · doi:10.1016/j.neuroimage.2025.121243 · Cited by 4
Closed-loop motor imagery EEG simulation for brain-computer interfaces
Shin, H., Suma, D., & He, B. (2022) · Frontiers in Human Neuroscience · doi:10.3389/fnhum.2022.951591 · Cited by 4
“…are the subject of active research due to their potential as neural prosthetic systems capable of improving quality of life for patients suffering from various motor function impairing conditions such as spinal cord injury, amyotrophic lateral sclerosis, and stroke (Anderson, 2004; He et al., 2020).”
BackgroundAn adaptive cross-class transfer learning framework with two-level alignment
Xu, D., Sun, Y.-J., & Li, M. (2023) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2023.105155 · Cited by 3
Front-End Replication Dynamic Window (FRDW) for Online Motor Imagery Classification
Chen, X., An, J., Wu, H., Li, S., Liu, B., & Wu, D. (2023) · IEEE Transactions on Neural Systems and Rehabilitation Engineering · doi:10.1109/tnsre.2023.3321640 · Cited by 5
“3) We integrate Euclidean alignment (EA) [15] with FRDW to accommodate individual differences in online cross-subject MI classification.”
BackgroundMethodSelf-citationCan Inter-Subject Associativity Predict Data-Driven BCI Performance?
Saha, S., Baumert, M., & McEwan, A. L. (2023) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc40787.2023.10340490 · Cited by 6
“Studies addressed the covariate shift primarily by data-driven transfer learning strategies [7]–[11].”
BackgroundMethodAdaptive Fusion and Transfer Learning for Enhanced E –Commerce Recommendations
Bagga, V., Sugunan, S., Srivastava, A., Kumar, R., Gupta, P. A., Kumar, D., & Guha, D. D. (2023) · Procedia Computer Science · doi:10.1016/j.procs.2023.12.037 · Cited by 4
Wang, X., Li, B., Lin, Y., & Gao, X. (2024) · Journal of Neural Engineering · doi:10.1088/1741-2552/ad2710 · Cited by 3
“The MDA-TSC was compared with several state-of-the-art methods, including EA + xDAWN [20], XGB-DIM [15], CORAL [30], EEGNet [16], EEG-Inception [18], EEGNet-adv [23] and MFSAN [25].”
InfluentialLuo, T.-J. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-024-18648-4 · Cited by 4
“(4) CA-CSP-TCA [29]: Based on the CA-CSP method, before classified by the LDA classifier, the CSP features are domain adapted by the TCA method.”
InfluentialPrivacy-Preserving Attention-Weighted Multi-Source Domain Adaptation for EEG Motor Imagery
Huang, Y., Hung, H.-N., & Tseng, V. S. M. (2024) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp48485.2024.10447934 · Cited by 3
“Despite existing works in MI UDA [7] [8], exploring varying source importance remains necessary.”
BackgroundMethodXu, Z., Cui, W., & Yang, L. (2024) · International Conference on Industrial Automation, Robotics and Control Engineering (IARCE) · doi:10.1109/iarce64300.2024.00071 · Cited by 3
“As an example, He et al. [11] introduced a data alignment framework in Euclidean space, aligning covariance matrices across various subjects.”
Morales, Ó. W. G., Escalante-Escobar, S., Collazos-Huertas, D. F., Álvarez-Meza, A. M., & Castellanos-Dominguez, G. (2025) · Applied Sciences · doi:10.3390/app15148036 · Cited by 7
Across-animal odor decoding by probabilistic manifold alignmentPreprint
Herrero-Vidal, P., Rinberg, D. A., & Savin, C. (2021) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2021.06.06.447279 · Cited by 9
“Lastly, a common decoder is trained on the aligned across-condition data [21, 9, 8].”
BackgroundMethodAhmed, S. G., Ambali Parambil, M. M., Damseh, R., Bouktif, S., Alnajjar, F., & Belkacem, A. N. (2026) · Frontiers in Systems Neuroscience · doi:10.3389/fnsys.2026.1840121 · Cited by 2
“Alternatively, He and Wu (2019) proposed an effective Euclidean Alignment (EA) approach that directly aligns and transforms raw EEG trials in the Euclidean space, with promising use cases for transfer learning in MI and ERP tasks.”
Huang, H.-Y., & Lin, Y. (2023) · Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society · doi:10.1109/embc40787.2023.10340188 · Cited by 2
“Future efforts, such as inter-session data alignment [16] and adaptive ensemble modeling [17], can augment capability of TL. Existing TL approaches have commonly been validated on the limited cross-day sessions of an individual [8-10 , 15”
Method
Metadata
- Authors
- He, H., & Wu, D.
- Affiliations
- Huazhong University of Science and Technology (CN)
- ORCID
- Venue
- IEEE Transactions on Biomedical Engineering
- Published
- 2020-02
- Volume
- 67
- Issue
- 2
- Pages
- 399-410
- ISSN
- 0018-9294 (ISSN-L)
- 1558-2531
- OpenAlex
- W2963727766
- Cited by
- 569
- References
- 52
- Citations by year
- 2026: 135 · 2025: 117 · 2024: 101 · 2023: 69 · 2022: 70 · 2021: 39 · 2020: 26
- FWCI
- 19.17
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Licence
- Locations
- IEEE Transactions on Biomedical Engineering (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- arXiv (Cornell University)
- Topics
- Life Sciences › Neuroscience › Cognitive Neuroscience › EEG and Brain-Computer Interfaces
- Physical Sciences › Computer Science › Artificial Intelligence › Emotion and Mood Recognition
- Physical Sciences › Computer Science › Human-Computer Interaction › Gaze Tracking and Assistive Technology
- Keywords
- brain-computer interface, EEG, transfer learning, data alignment, interindividual variability, motor imagery classification, unsupervised alignment, EEG preprocessing
- MeSH terms
- Machine Learning, Algorithms, Computer Simulation, Electroencephalography, Evoked Potentials, Humans, Imagination, Signal Processing, Computer-Assisted, Databases, Factual, Brain-Computer Interfaces
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (gpt-6-sol).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.