SpecMoE: Spectral Mixture-of-Experts Foundation Model for Cross-Species EEG Decoding

TL;DR

This paper introduces SpecMoE, a novel spectral mixture-of-experts foundation model for EEG decoding that employs a Gaussian-smoothed masking scheme on STFT maps, achieving state-of-the-art results across multiple tasks and species.

Contribution

It proposes a new masking strategy and a hierarchical architecture for EEG decoding, along with a mixture of experts framework for large-scale pretraining and cross-species generalization.

Findings

Achieves state-of-the-art performance on sleep staging, emotion recognition, and other EEG tasks.

Demonstrates strong cross-species and cross-subject generalization.

Outperforms existing models in diverse EEG decoding scenarios.

Abstract

Decoding the orchestration of neural activity in electroencephalography (EEG) signals is a central challenge in bridging neuroscience with artificial intelligence. Foundation models have made strides in generalized EEG decoding, yet many existing frameworks primarily relying on separate temporal and spectral masking of raw signals during self-supervised pretraining. Such strategies often tend to bias learning toward high-frequency oscillations, as low-frequency rhythmic patterns can be easily inferred from the unmasked signal. We introduce a foundation model that utilizes a novel Gaussian-smoothed masking scheme applied to short-time Fourier transform (STFT) maps. By jointly applying time, frequency, and time-frequency Gaussian masks, we make the reconstruction task much more challenging, forcing the model to learn intricate neural patterns across both high- and low-frequency domains. To effectively recover signals under this aggressive masking strategy, we design SpecHi-Net, a U-shaped hierarchical architecture with multiple encoding and decoding stages. To accelerate large-scale pretraining, we partition the data into three subsets, each used to train an independent expert model. We then combine these models through SpecMoE, a mixture of experts framework guided by a learned spectral gating mechanism. SpecMoE achieves state-of-the-art performance across a diverse set of EEG decoding tasks, including sleep staging, emotion recognition, motor imagery classification, abnormal signal detection, and drug effect prediction. Importantly, the model demonstrates strong cross-species and cross-subject generalization, maintaining high accuracy on both human and murine EEG datasets.