按频段分治的脑电基础模型，提升跨任务迁移能力
BandVQ: Band-Wise Vector-Quantized EEG Foundation Model
- 将脑电信号分解为5个频段，每段独立训练向量量化编码器
- 在超9000名受试者数据上预训练，6个分类任务均达领先或持平表现
- 适合需要高精度脑电特征提取的研究者，尤其关注认知与运动想象任务
脑电基础建模的核心挑战在于学习跨不同任务、导联配置和频谱特性的可迁移表示。现有掩码建模方法多依赖宽带连续块或单一离散表示，可能忽略频段特异性活动。本文提出BandVQ，一种基于频段的向量量化脑电基础模型：将脑电信号分解为δ、θ、α、β、γ五个频段，为每个频段独立训练VQ-VAE编码器，并在生成的离散代码索引上预训练共享Transformer编码器。编码器使用掩码代码、量化绝对对数功率、通道与时间嵌入，以及代表参考、频段、任务类型和阶段的元数据前缀。引入区域掩码以降低空间相邻电极的平凡重建。模型在71个公开脑电数据集上预训练，覆盖超9200名受试者、35.7万小时单通道数据，在6个无个体依赖的分类任务上评估。当前设置下，该模型在三个认知任务上达到最高报告性能，在三个运动想象任务上表现具有竞争力。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。