BrainRVQ: A High-Fidelity EEG Foundation Model via Dual-Domain Residual Quantization and Hierarchical Autoregression

Mingzhe Cui; Tao Chen; Yang Jiao; Yiqin Wang; Lei Xie; Yi Pan; Luca Mainardi

eess.SPFeb 18, 2026

BrainRVQ: A High-Fidelity EEG Foundation Model via Dual-Domain Residual Quantization and Hierarchical Autoregression

Authors: Mingzhe Cui, Tao Chen, Yang Jiao, Yiqin Wang, Lei Xie, Yi Pan, Luca Mainardi

Organizations: State Key Laboratory of Industrial Control Technology, Zhejiang University, Hangzhou, China · Department of Electronics, Information and Bioengineering, Politecnico di Milano, Milan, Italy · Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences, Shenzhen, China

Abstract

Developing foundation models for electroencephalography (EEG) remains challenging due to the signal's low signal-to-noise ratio and complex spectro-temporal non-stationarity. Existing approaches often overlook the hierarchical latent structure inherent in neural dynamics, leading to suboptimal reconstruction of fine-grained information. In this work, we propose BrainRVQ, a general-purpose EEG foundation model pre-trained on a large-scale corpus of clinical EEG data. Unlike standard masked modeling, BrainRVQ features a Dual-Domain Residual Vector Quantization (DD-RVQ) tokenizer that disentangles temporal waveforms and spectral patterns into hierarchical discrete codes. We further introduce a hierarchical autoregressive pre-training objective that learns to reconstruct these codes in a coarse-to-fine manner, utilizing an importance-guided curriculum masking strategy to prioritize information-rich neural events over background noise. Extensive experiments across 8 diverse downstream datasets demonstrate that BrainRVQ consistently outperforms state-of-the-art baselines, validating its effectiveness in learning robust and generalizable neural representations. Our code and model weights are available:https://github.com/keqicmz/BrainRVQ

Figures & tables

Figure 1: The overall architecture of the BrainRVQ framework. (A) DDRVQ for EEG Tokenization: The module employs a DDRVQ mechanism to discretize EEG signals. It extracts features simultaneously in time and frequency domains and is optimized via a joint objective of reconstruction and embedding loss. (B) Pre-training Stage: We introduce a Hierarchical Autoregressive Masked Modeling objective. The model learns to predict residual tokens in a coarse-to-fine manner, guided by an Importance-Guided Curriculum Masking strategy that prioritizes high-information regions. (C) Downstream Adaptation: The pre-trained encoder serves as a general-purpose feature extractor.

Task	Dataset	#Ch	Win(s)	#Samps	Type
Abnormal Det.	TUAB ( Obeid and Picone, 2016 )	16	10	409,455	Binary
Event Class.	TUEV ( Obeid and Picone, 2016 )	16	5	112,491	6-Class
Seizure Det.	CHB-MIT ( Shoeb, 2009 )	16	10	326,993	Binary
Emotion Rec.	SEED-V ( Liu et al., 2021 )	62	1	117,744	5-Class
Motor Imagery	PhysioNet ( Schalk et al., 2004 )	64	4	9,837	4-Class
	SHU-MI ( Ma et al., )	32	4	11,988	Binary

Table 1: Overview of the 8 downstream datasets across diverse BCI tasks.

Method	Mental Workload			CHB-MIT
Method	Bal. Acc	AUC-PR	AUROC	Bal. Acc	AUC-PR	AUROC
EEGNet	0.677 $\pm$ 0.012	0.576 $\pm$ 0.010	0.732 $\pm$ 0.011	0.566 $\pm$ 0.011	0.191 $\pm$ 0.018	0.805 $\pm$ 0.014
ST-Transformer	0.663 $\pm$ 0.017	0.567 $\pm$ 0.026	0.713 $\pm$ 0.017	0.592 $\pm$ 0.020	0.142 $\pm$ 0.009	0.824 $\pm$ 0.049
BENDR	0.568 $\pm$ 0.045	0.366 $\pm$ 0.067	0.568 $\pm$ 0.045	0.561 $\pm$ 0.043	0.307 $\pm$ 0.124	0.863 $\pm$ 0.053
BIOT	0.688 $\pm$ 0.019	0.600 $\pm$ 0.020	0.754 $\pm$ 0.014	0.707 $\pm$ 0.046	0.328 $\pm$ 0.046	0.876 $\pm$ 0.028
LaBraM	0.691 $\pm$ 0.013	0.600 $\pm$ 0.016	0.772 $\pm$ 0.009	0.708 $\pm$ 0.036	0.329 $\pm$ 0.040	0.868 $\pm$ 0.020

Table 2: Main evaluation results on four representative downstream tasks. The table is divided into binary classification (top) and multi-class classification (bottom). Best results are in bold , second best are underlined .

(a) Dual-Domain Token.
Configuration	Binary (AUROC $\uparrow$ )		Multi-class (Kappa $\uparrow$ )
Configuration	Mental	CHB-MIT	TUEV	BCICIV-2a
Full Model	0.862	0.928	0.690	0.388
w/o Freq codebook	0.845	0.887	0.639	0.373
w/o Time codebook	0.857	0.896	0.655	0.374
(b) Residual Quant.

Table 3: Ablation study results. AUROC for binary and Cohen’s Kappa for multi-class tasks.

Figure 2: Visualization of ablation study results. The proposed Full Model (dark blue) consistently outperforms single-domain and non-hierarchical variants across all datasets. Left: AUROC scores for Mental Workload and CHB-MIT. Right: Cohen’s Kappa scores for TUEV and BCICIV-2a.

Appendix figures & tables28 assets

Supplementary material from the paper’s appendix.

Appendix

EEG Sample
Hyperparameters	Tokenizer	Pre-training
Channels	19
Time points	6000 (30 seconds @ 200 Hz)
Patch dimension	200 (1 second)
Sequence length	$19\times 30=570$
Mask ratio	–	0.5

Appendix

Table 4: Hyperparameters for BrainRVQ pre-training.

Figure 3: DD-RVQ tokenizer training dynamics. The total loss (black) comprises time-domain waveform reconstruction (blue), frequency-domain amplitude reconstruction (orange), and phase reconstruction (green).

Figure 4: Pre-training loss dynamics. Layer-0 (coarse) achieves consistently lower loss than Layer-1/2 (fine), validating the coarse-to-fine learning hierarchy.

Figure 5: Layer-wise token prediction accuracy during pre-training. Layer-0 achieves the highest accuracy (

\sim

29%), followed by Layer-1 (

\sim

22%) and Layer-2 (

\sim

14%), confirming that coarse codes are more predictable than fine-grained residuals.

Figure 6: Number of unused codes during DD-RVQ training for (a) temporal and (b) frequency codebooks. All layers converge to near-zero unused codes within 20 epochs, achieving 100% codebook utilization.

Figure 7: Final codebook usage distribution for (a) temporal and (b) frequency domains across three RVQ layers. All codebooks achieve 100% utilization with near-uniform distributions (normalized entropy

>

0.99).

Codebook	Norm. Entropy	Gini Coef.	Top-10% Contrib.
Temporal Layer-0	0.994	0.174	18.1%
Temporal Layer-1	0.996	0.151	15.4%
Temporal Layer-2	0.998	0.104	13.8%
Frequency Layer-0	0.994	0.174	18.1%
Frequency Layer-1	0.996	0.151	15.5%
Frequency Layer-2	0.998	0.104	13.7%

Appendix

Table 5: Codebook utilization statistics. Normalized entropy close to 1.0 and low Gini coefficients indicate near-uniform code usage across all layers.

Hyperparameter	Value
Optimizer	AdamW
Adam $\beta$	(0.9, 0.999)
Adam $\epsilon$	$1\times 10^{-8}$
Learning rate	$5\times 10^{-4}$
Weight decay	$5\times 10^{-2}$
Layer decay	0.65

Appendix

Table 6: Fine-tuning hyperparameters for downstream tasks.

Method	Bal. Acc	AUC-PR	AUROC
EEGNet	0.589 $\pm$ 0.018	0.631 $\pm$ 0.014	0.628 $\pm$ 0.015
ST-Transformer	0.599 $\pm$ 0.021	0.639 $\pm$ 0.012	0.643 $\pm$ 0.011
BENDR	0.557 $\pm$ 0.023	0.585 $\pm$ 0.027	0.586 $\pm$ 0.028
BIOT	0.618 $\pm$ 0.018	0.677 $\pm$ 0.012	0.661 $\pm$ 0.013
LaBraM	0.617 $\pm$ 0.019	0.676 $\pm$ 0.008	0.660 $\pm$ 0.009
CBraMod	0.637 $\pm$ 0.015	0.714 $\pm$ 0.009	0.699 $\pm$ 0.007

Appendix

Table 7: Results on motor imagery classification (SHU-MI, 2-class). Best results are in bold , second best are underlined .

Method	Bal. Acc	Kappa	W-F1
EEGNet	0.581 $\pm$ 0.013	0.447 $\pm$ 0.020	0.580 $\pm$ 0.012
ST-Transformer	0.604 $\pm$ 0.008	0.471 $\pm$ 0.020	0.605 $\pm$ 0.008
BENDR	0.422 $\pm$ 0.022	0.401 $\pm$ 0.025	0.408 $\pm$ 0.024
BIOT	0.615 $\pm$ 0.015	0.488 $\pm$ 0.027	0.616 $\pm$ 0.020
LaBraM	0.617 $\pm$ 0.012	0.491 $\pm$ 0.019	0.618 $\pm$ 0.014
CBraMod	0.642 $\pm$ 0.009	0.522 $\pm$ 0.017	0.643 $\pm$ 0.010

Appendix

Table 8: Results on motor imagery classification (PhysioNet-MI, 4-class). Best results are in bold , second best are underlined .

Method	Bal. Acc	Kappa	W-F1
EEGNet	0.296 $\pm$ 0.010	0.101 $\pm$ 0.014	0.275 $\pm$ 0.010
ST-Transformer	0.305 $\pm$ 0.007	0.108 $\pm$ 0.012	0.283 $\pm$ 0.011
BENDR	0.223 $\pm$ 0.006	0.034 $\pm$ 0.006	0.203 $\pm$ 0.033
BIOT	0.384 $\pm$ 0.019	0.226 $\pm$ 0.026	0.386 $\pm$ 0.020
LaBraM	0.398 $\pm$ 0.014	0.239 $\pm$ 0.021	0.397 $\pm$ 0.011
CBraMod	0.409 $\pm$ 0.010	0.257 $\pm$ 0.014	0.410 $\pm$ 0.011

Appendix

Table 9: Results on emotion recognition (SEED-V, 5-class). Best results are in bold , second best are underlined .

Method	Bal. Acc	AUC-PR	AUROC
EEGNet	0.764 $\pm$ 0.004	0.830 $\pm$ 0.004	0.841 $\pm$ 0.003
ST-Transformer	0.797 $\pm$ 0.002	0.852 $\pm$ 0.003	0.871 $\pm$ 0.002
BENDR	0.771 $\pm$ 0.025	0.841 $\pm$ 0.022	0.843 $\pm$ 0.024
BIOT	0.796 $\pm$ 0.006	0.879 $\pm$ 0.002	0.882 $\pm$ 0.004
LaBraM	0.814 $\pm$ 0.002	0.897 $\pm$ 0.002	0.902 $\pm$ 0.001
CBraMod	0.829 $\pm$ 0.002	0.926 $\pm$ 0.001	0.923 $\pm$ 0.001

Appendix

Table 10: Results on abnormal detection (TUAB, 2-class). Best results are in bold , second best are underlined .

	Mental Workload, 2-class			BCICIV-2a, 4-class
RVQ Layers	Bal. Acc	AUC-PR	AUROC	Bal. Acc	Kappa	W-F1
1	0.740 $\pm$ 0.010	0.756 $\pm$ 0.010	0.860 $\pm$ 0.012	0.491 $\pm$ 0.011	0.322 $\pm$ 0.008	0.480 $\pm$ 0.013
2	0.741 $\pm$ 0.015	0.761 $\pm$ 0.013	0.863 $\pm$ 0.010	0.534 $\pm$ 0.015	0.383 $\pm$ 0.010	0.551 $\pm$ 0.016
3 (Ours)	0.747 $\pm$ 0.011	0.758 $\pm$ 0.012	0.862 $\pm$ 0.010	0.541 $\pm$ 0.008	0.388 $\pm$ 0.008	0.533 $\pm$ 0.012
5	0.724 $\pm$ 0.016	0.721 $\pm$ 0.018	0.816 $\pm$ 0.014	0.536 $\pm$ 0.014	0.382 $\pm$ 0.013	0.528 $\pm$ 0.015
7	0.701 $\pm$ 0.012	0.692 $\pm$ 0.018	0.824 $\pm$ 0.016	0.503 $\pm$ 0.012	0.337 $\pm$ 0.014	0.489 $\pm$ 0.013

Appendix

Table 11: Ablation study on RVQ depth. Best results are in bold , second best are underlined .

Mask Ratio	Balanced Accuracy	AUC-PR	AUROC
0.2	0.718 $\pm$ 0.010	0.725 $\pm$ 0.014	0.831 $\pm$ 0.012
0.3	0.734 $\pm$ 0.012	0.742 $\pm$ 0.012	0.848 $\pm$ 0.011
0.4	0.744 $\pm$ 0.009	0.754 $\pm$ 0.012	0.859 $\pm$ 0.010
0.5	0.747 $\pm$ 0.011	0.758 $\pm$ 0.012	0.862 $\pm$ 0.010
0.6	0.741 $\pm$ 0.010	0.758 $\pm$ 0.013	0.857 $\pm$ 0.013
0.7	0.730 $\pm$ 0.011	0.755 $\pm$ 0.014	0.850 $\pm$ 0.012

Appendix

Table 12: Performance of BrainRVQ on Mental Workload Dataset under Different Mask Ratios.

Mask Ratio	Balanced Accuracy	Cohen’s Kappa	Weighted F1
0.2	0.509 $\pm$ 0.010	0.352 $\pm$ 0.010	0.501 $\pm$ 0.014
0.3	0.526 $\pm$ 0.011	0.371 $\pm$ 0.012	0.518 $\pm$ 0.013
0.4	0.537 $\pm$ 0.008	0.384 $\pm$ 0.008	0.529 $\pm$ 0.012
0.5	0.541 $\pm$ 0.010	0.388 $\pm$ 0.008	0.533 $\pm$ 0.012
0.6	0.540 $\pm$ 0.012	0.390 $\pm$ 0.010	0.534 $\pm$ 0.012
0.7	0.522 $\pm$ 0.009	0.368 $\pm$ 0.010	0.515 $\pm$ 0.013

Appendix

Table 13: Performance of BrainRVQ on BCICIV-2a Dataset under Different Mask Ratios.

Figure 8: Impact of masking ratio on downstream performance across six metrics. Top Row: Mental Workload dataset metrics (Balanced Accuracy, AUC-PR, AUROC). Bottom Row: BCICIV-2a dataset metrics (Balanced Accuracy, Cohen’s Kappa, Weighted F1). Performance consistently peaks at moderate masking ratios (0.5), validating the inverted U-shaped trend. Shaded regions represent standard deviation.

Figure 9: Reconstruction examples from the DD-RVQ tokenizer. Each row shows a different EEG segment. (a) Time-domain waveform reconstruction. (b) Amplitude spectrum reconstruction. (c) Phase spectrum reconstruction. Blue: original signal; Orange dashed: reconstructed signal. Pearson correlation coefficients (

r

) are displayed for each panel.

Metric	Time Domain	Amplitude	Phase
MSE ( $\times 10^{-3}$ )	178.62 $\pm$ 50.63	64.49 $\pm$ 17.51	653.32 $\pm$ 49.92
Correlation	0.904 $\pm$ 0.030	0.956 $\pm$ 0.016	0.577 $\pm$ 0.042
SNR (dB)	8.1 $\pm$ 1.2	11.2 $\pm$ 1.2	2.0 $\pm$ 0.5

Appendix

Table 14: Reconstruction quality metrics averaged over 7,928 EEG samples.

Figure 10: Reconstruction quality with different RVQ depths. Left column: reconstructed waveforms (colored dashed) overlaid on the original signal (gray solid). Right column: residual errors. Correlation coefficients, MSE, and SNR are displayed for each configuration.

RVQ Layers	MSE ( $\times 10^{-3}$ )	Correlation	SNR (dB)
1 layer	348.69	0.807	4.6
2 layers	224.59	0.880	6.5
3 layers	157.18	0.918	8.0
5 layers	129.82	0.932	8.8
7 layers	139.04	0.928	8.5

Appendix

Table 15: Reconstruction quality metrics for different RVQ depths.

Figure 11: Comparison of mask patterns between random and importance-guided masking. Shaded regions indicate masked patches, with green highlighting high-information patches (top 30% by information score). Dashed boxes mark high-information patches that remain visible. (a) Random masking distributes masks uniformly, often missing informative regions. (b) Importance-guided masking preferentially targets high-information patches, ensuring the model learns to reconstruct complex neural dynamics.

Figure 12: Information score heatmap across patches. Higher scores (darker green) indicate patches with greater neural band power (4–30 Hz), lower artifact contamination, and higher Hjorth complexity. White borders highlight patches in the top 30% of information scores. The heterogeneous distribution demonstrates that neural information is sparsely distributed across EEG recordings.

Figure 13: Distribution of information scores for masked vs. visible patches. (a) Under random masking, both distributions are nearly identical. (b) Under importance-guided masking, masked patches have significantly higher mean information scores, confirming that our strategy systematically targets informative regions.

Data Volume	Balanced Accuracy	AUC-PR	AUROC
1k hours	0.695 $\pm$ 0.012	0.702 $\pm$ 0.013	0.810 $\pm$ 0.010
2k hours	0.710 $\pm$ 0.013	0.718 $\pm$ 0.015	0.825 $\pm$ 0.012
3k hours	0.722 $\pm$ 0.008	0.730 $\pm$ 0.014	0.838 $\pm$ 0.013
5k hours	0.732 $\pm$ 0.011	0.741 $\pm$ 0.010	0.848 $\pm$ 0.005
7k hours	0.740 $\pm$ 0.010	0.750 $\pm$ 0.012	0.856 $\pm$ 0.011
9k hours	0.747 $\pm$ 0.011	0.758 $\pm$ 0.012	0.862 $\pm$ 0.010

Appendix

Table 16: Data Scaling on Mental Workload Dataset (Binary Classification).

Data Volume	Balanced Accuracy	Cohen’s Kappa	Weighted F1
1k hours	0.485 $\pm$ 0.010	0.318 $\pm$ 0.015	0.478 $\pm$ 0.016
2k hours	0.502 $\pm$ 0.010	0.338 $\pm$ 0.015	0.495 $\pm$ 0.011
3k hours	0.515 $\pm$ 0.011	0.354 $\pm$ 0.020	0.508 $\pm$ 0.013
5k hours	0.525 $\pm$ 0.009	0.368 $\pm$ 0.011	0.518 $\pm$ 0.008
7k hours	0.534 $\pm$ 0.009	0.379 $\pm$ 0.005	0.527 $\pm$ 0.015
9k hours	0.541 $\pm$ 0.008	0.388 $\pm$ 0.008	0.533 $\pm$ 0.012

Appendix

Table 17: Data Scaling on BCICIV-2a Dataset (4-Class Motor Imagery).

Layers	Parameters	Balanced Accuracy	AUC-PR	AUROC
4	1.96M	0.720 $\pm$ 0.012	0.728 $\pm$ 0.014	0.835 $\pm$ 0.013
6	2.93M	0.730 $\pm$ 0.011	0.738 $\pm$ 0.010	0.845 $\pm$ 0.015
8	3.89M	0.738 $\pm$ 0.015	0.747 $\pm$ 0.016	0.853 $\pm$ 0.011
10	4.86M	0.743 $\pm$ 0.023	0.753 $\pm$ 0.018	0.858 $\pm$ 0.018
12	5.82M	0.747 $\pm$ 0.011	0.758 $\pm$ 0.012	0.862 $\pm$ 0.010

Appendix

Table 18: Model Scaling on Mental Workload Dataset (Binary Classification).

Layers	Parameters	Balanced Accuracy	Cohen’s Kappa	Weighted F1
4	1.96M	0.508 $\pm$ 0.012	0.345 $\pm$ 0.013	0.501 $\pm$ 0.014
6	2.93M	0.520 $\pm$ 0.006	0.360 $\pm$ 0.008	0.513 $\pm$ 0.010
8	3.89M	0.530 $\pm$ 0.015	0.372 $\pm$ 0.010	0.523 $\pm$ 0.012
10	4.86M	0.536 $\pm$ 0.015	0.381 $\pm$ 0.012	0.529 $\pm$ 0.015
12	5.82M	0.541 $\pm$ 0.008	0.388 $\pm$ 0.008	0.533 $\pm$ 0.012

Appendix

Table 19: Model Scaling on BCICIV-2a Dataset (4-Class Motor Imagery).

	SEED-V, 5-class			SHU-MI, 2-class
Methods	Bal. Acc	Kappa	W-F1	Bal. Acc	AUC-PR	AUROC
CBraMod (full data)	0.409 $\pm$ 0.010	0.257 $\pm$ 0.014	0.410 $\pm$ 0.011	0.637 $\pm$ 0.015	0.714 $\pm$ 0.009	0.699 $\pm$ 0.007
Ours (full data)	0.407 $\pm$ 0.008	0.260 $\pm$ 0.010	0.414 $\pm$ 0.011	0.637 $\pm$ 0.018	0.723 $\pm$ 0.011	0.717 $\pm$ 0.007
BIOT (30%)	0.351 $\pm$ 0.038	0.178 $\pm$ 0.043	0.349 $\pm$ 0.042	0.542 $\pm$ 0.035	0.626 $\pm$ 0.027	0.602 $\pm$ 0.030
LaBraM (30%)	0.369 $\pm$ 0.031	0.204 $\pm$ 0.038	0.370 $\pm$ 0.032	0.564 $\pm$ 0.029	0.652 $\pm$ 0.023	0.647 $\pm$ 0.027
CBraMod (30%)	0.388 $\pm$ 0.024	0.229 $\pm$ 0.025	0.389 $\pm$ 0.026	0.623 $\pm$ 0.020	0.690 $\pm$ 0.013	0.675 $\pm$ 0.011

Appendix

Table 20: Performance comparison on low-resource settings with 30% of fine-tuning data.

EEG Foundation Models

Electroencephalography (EEG) foundation models learn reusable representations of brain signals for transfer across people, tasks, and recording setups.

See all 100 papers

BandVQ: Band-Wise Vector-Quantized EEG Foundation Model
May 24, 2026Jamiyan Sukhbaatar, Satoshi Imamura, Toshihisa TanakaCross-Subject EEG Decoding Masked Autoencoders
National University of Mongolia, Ulaanbaatar, Mongolia · Tokyo University of Agriculture and Technology, Tokyo, Japan
MSBraM: A Multi-scale Self-supervised Brain Foundation Model for Hierarchical EEG Dynamics Learning
Jul 23, 2026Tao Zhou, Jing Han, Lingyu Shu +1Electroencephalography Self-Supervised Pre-Training
Hunan University Changsha, China
Standing on the Shoulders of Giants: Rethinking EEG Foundation Model Pretraining via Multi-Teacher Distillation
Mar 4, 2026Chenqi Li, Yu Liu, Shuo Zhang +2EEG Foundation Models Multi-Teacher Knowledge Distillation
Department of Engineering Science, University of Oxford.

May 24, 2026J/K move · Enter open · S save

BandVQ: Band-Wise Vector-Quantized EEG Foundation Model

Jamiyan Sukhbaatar, Satoshi Imamura, Toshihisa Tanaka

National University of Mongolia, Ulaanbaatar, Mongolia · Tokyo University of Agriculture and Technology, Tokyo, Japan

Cross-Subject EEG Decoding Masked Autoencoders EEG Decoding Frequency-Domain Feature Learning Vector-Quantized Variational Autoencoder EEG Foundation Models

Open paper AskPDF arXiv

A central challenge in electroencephalography (EEG) foundation modeling is learning transferable representations across recordings with diverse tasks, montages, references, and spectral characteristics. Existing masked modeling approaches often rely on broadband continuous patches or a single discrete representation, which may underrepresent frequency-specific activity. This paper proposes BandVQ, a band-wise vector-quantized EEG foundation model that decomposes EEG into delta, theta, alpha, beta, and gamma bands, trains an independent VQ-VAE tokenizer for each band, and pretrains a shared Transformer encoder on the resulting discrete VQ code indices. The encoder uses masked code tokens, quantized absolute log-power tokens, channel and temporal embeddings, and metadata prefix tokens representing reference, band, task family, and phase. Region-based masking is also introduced to reduce the trivial reconstruction of spatially adjacent electrodes. The model is pretrained on 71 public EEG corpora comprising over 9,200 subjects and 357,000 single-channel hours and evaluated on six subject-independent classification datasets. Under the current evaluation setting, the proposed model achieves strong transfer performance, with the highest reported results on three cognitive tasks and competitive performance on three motor imagery tasks.

Abstract

EEG Foundation Models

Explore similar work

BandVQ: Band-Wise Vector-Quantized EEG Foundation Model

MSBraM: A Multi-scale Self-supervised Brain Foundation Model for Hierarchical EEG Dynamics Learning

Standing on the Shoulders of Giants: Rethinking EEG Foundation Model Pretraining via Multi-Teacher Distillation