![[2602.16951] BrainRVQ: A High-Fidelity EEG Foundation Model via Dual-Domain Residual Quantization and Hierarchical Autoregression](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.16951] BrainRVQ: A High-Fidelity EEG Foundation Model via Dual-Domain Residual Quantization and Hierarchical Autoregression
Summary
The paper presents BrainRVQ, a high-fidelity EEG foundation model that utilizes Dual-Domain Residual Quantization and Hierarchical Autoregression to enhance the reconstruction of EEG signals, outperforming existing methods across various datasets.
Why It Matters
This research addresses the challenges in EEG signal processing, particularly the low signal-to-noise ratio and complex dynamics. By introducing a novel model that effectively captures hierarchical structures in neural data, it has significant implications for clinical applications and neuroscience research.
Key Takeaways
- BrainRVQ improves EEG signal reconstruction through advanced modeling techniques.
- The model utilizes a unique Dual-Domain Residual Vector Quantization approach.
- Hierarchical autoregression allows for better handling of neural dynamics.
- Extensive testing shows superior performance compared to existing models.
- The research contributes to the development of robust EEG analysis tools.
Electrical Engineering and Systems Science > Signal Processing arXiv:2602.16951 (eess) [Submitted on 18 Feb 2026] Title: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 View a PDF of the paper titled BrainRVQ: A High-Fidelity EEG Foundation Model via Dual-Domain Residual Quantization and Hierarchical Autoregression, by Mingzhe Cui and 6 other authors View PDF HTML (experimental) 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. Extensiv...