ScaleBrain: A Scale-Consistent Framework for EEG Foundation Models
Abstract
Foundation models pre-trained on large unlabeled electroencephalography (EEG) corpora have emerged as a transformative paradigm for Brain-Computer Interface (BCI) applications. Recent EEG foundation models have shown that multi-scale representations derived from fixed-length patches can improve EEG representation learning. Existing frameworks commonly construct temporal scales through feature-level operations. However, these operations leave the correspondence between learned features and scale-specific characteristics of the original EEG signals implicit. Moreover, when the self-supervised objective is not aligned with the multi-scale input representations, information at some temporal scales may receive insufficient supervision during pre-training. In light of this, we propose ScaleBrain, a scale-consistent pre-training framework that explicitly coordinates multi-scale representations of EEG from both the model input and optimization objective sides. Specifically, ScaleBrain introduces multi-scale EEG tokenization to represent the same EEG segment through analysis windows with different temporal spans and shared temporal anchors, and pairs these representations with scale-specific reconstruction to align with the multi-scale input structure and provide supervision across temporal resolutions during self-supervised learning. We further develop an axial spatio-temporal attention mechanism to contextualize the fused multi-scale representation along spatial and temporal dimensions. Extensive evaluations on six public BCI benchmarks against 12 representative baselines show that ScaleBrain achieves the highest mean Balanced Accuracy on five datasets and ranks third on the remaining one, with only 5.0M parameters, demonstrating strong performance and favorable parameter efficiency.
Related
- Wetl-SAM2YNet: A foundation model driven framework for fine-scale wetland mapping
- A scale-consistent 0.1° dataset of forest productivity and environmental drivers for China (1990–2018)
- Re-evaluating the benefit of pretraining scale in cell foundation models
- ALHENA, an H&E encoder distilled from transcriptomic foundation models for multi-scale slide analysis
- Learning from tandem mass spectra at scale with a self-supervised foundation model for proteomics
BibTeX
@misc{zhou2026scalebrain,
title = {{ScaleBrain: A Scale-Consistent Framework for EEG Foundation Models}},
author = {Minyu Zhou and Yajing Wu and Yongqiang Tang and Wang Jia and Wensheng Zhang},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.14035},
eprint = {2611.14035},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.14035},
}
RIS
TY - UNPB
TI - ScaleBrain: A Scale-Consistent Framework for EEG Foundation Models
AU - Minyu Zhou
AU - Yajing Wu
AU - Yongqiang Tang
AU - Wang Jia
AU - Wensheng Zhang
PY - 2026
DA - 2026/11/10
PB - arXiv
DO - 10.48550/arxiv.2611.14035
UR - https://arxiv.org/abs/2611.14035
AB - Foundation models pre-trained on large unlabeled electroencephalography (EEG) corpora have emerged as a transformative paradigm for Brain-Computer Interface (BCI) applications. Recent EEG foundation models have shown that multi-scale representations derived from fixed-length patches can improve EEG representation learning. Existing frameworks commonly construct temporal scales through feature-level operations. However, these operations leave the correspondence between learned features and scale-specific characteristics of the original EEG signals implicit. Moreover, when the self-supervised objective is not aligned with the multi-scale input representations, information at some temporal scales may receive insufficient supervision during pre-training. In light of this, we propose ScaleBrain, a scale-consistent pre-training framework that explicitly coordinates multi-scale representations of EEG from both the model input and optimization objective sides. Specifically, ScaleBrain introduces multi-scale EEG tokenization to represent the same EEG segment through analysis windows with different temporal spans and shared temporal anchors, and pairs these representations with scale-specific reconstruction to align with the multi-scale input structure and provide supervision across temporal resolutions during self-supervised learning. We further develop an axial spatio-temporal attention mechanism to contextualize the fused multi-scale representation along spatial and temporal dimensions. Extensive evaluations on six public BCI benchmarks against 12 representative baselines show that ScaleBrain achieves the highest mean Balanced Accuracy on five datasets and ranks third on the remaining one, with only 5.0M parameters, demonstrating strong performance and favorable parameter efficiency.
ER -