Understanding and Correcting Low-Frequency Bias in EEG Foundation Model
2026-08-03 • Machine Learning
Machine Learning
AI summaryⓘ
The authors found that when training models to understand brain signals (EEG), the models tend to focus too much on slow, low-frequency parts of the signals and ignore other frequencies. This happens because of the natural properties of EEG signals and how the models learn using a common method that emphasizes bigger signal parts. To fix this, they created a new training method called FAME that treats all frequency bands equally, helping the model learn a more balanced understanding. Testing showed that FAME works better on many brain signal tasks, highlighting the need to balance learning across frequencies.
EEGfrequency bias1/f spectral structuremasked autoencoderL2 reconstruction losstime-frequency analysisspectral balancefoundation modelsOmniEEG-Bench
Authors
Junjie Yu, Zihan Deng, Jianyu Zhang, Junrong Mu, Jiahui An, Wenxiao Ma, Ziling Lu, Yue Wang, Yan Zhu, Kexin Lou, Quanying Liu
Abstract
Increasing EEG pretraining data scale or model capacity does not consistently improve downstream performance. We identify a persistent low-frequency bias in representations learned by diverse EEG foundation models, which remains across dataset scales, model capacities, and pretraining objectives. Our analysis links this bias to the interaction between EEG's $1/f^α$-like spectral structure and neural networks' tendency to preferentially learn low-frequency components. In masked autoencoders, the $\ell_2$ reconstruction objective further amplifies this imbalance: under comparable relative reconstruction errors, high-power low-frequency components contribute disproportionately to the loss. To address this issue, we introduce FAME, a frequency-balanced masked autoencoding framework that reconstructs time--frequency activity in predefined EEG bands from masked EEG inputs. FAME independently standardizes the reconstruction targets within each band and assigns equal weight to all band-specific losses, thereby balancing supervision across the EEG spectrum. Evaluated on 41 downstream tasks in OmniEEG-Bench, FAME learns more spectrally balanced representations and achieves state-of-the-art performance on 24 of them. These results underscore the importance of balanced spectral supervision for learning transferable EEG representations.