Enabling Unsupervised Training of Deep EEG Denoisers With Intelligent Partitioning

2026-05-07

AI Summary

This work addresses the challenge of denoising wearable electroencephalography (EEG) signals, which are typically weak, spectrally overlapping with noise, and lack clean reference signals. To overcome these limitations, the authors propose an intelligent partitioning self-supervised denoising method (iPSD), which intelligently splits a single noisy EEG segment into subsegments that share the same neural source but contain independent noise realizations. This strategy enables the construction of self-supervised training pairs, facilitating deep denoising without requiring clean labels for the first time. The method is particularly effective for ultra-low signal-to-noise ratio data (as low as –10 dB) acquired from wearable devices such as in-ear EEG systems, significantly outperforming existing approaches under strong electromyographic interference. Experimental results demonstrate orders-of-magnitude improvement in spectral fidelity and validate the method’s efficacy in real-world scenarios.

Abstract

Denoising wearable electroencephalogram (EEG) is inherently challenging since neural activity is not only subtle but also inseparable from spectrally overlapping noise artifacts. Classical signal processing methods, relying on fixed or heuristic rules, cannot handle the time-varying pervasive artifacts in wearable EEGs. Deep learning methods, on the other hand, show promise in decomposition-free EEG denoising using highly expressive neural networks, but the training requires artifact-free EEG, which is inherently unobtainable. To address this, we propose Intelligent Partitioning for Self-supervised Denoising (iPSD). Our method eliminates the need for clean references by learning to partition an input EEG segment into independent noisy realizations with the same underlying signal. This enables self-supervision of deep learning denoisers, even in zero-shot settings where only a single EEG segment to be denoised is available. We validate iPSD through extensive experiments, including validations on wearable EEG from in-ear sensors. The results show that iPSD achieves state-of-the-art performance, most notably under extremely low signal-to-noise ratios (down to -10 dB) and challenging artifacts (e.g., EMG), with spectral fidelity orders of magnitude higher than competitive baselines.

Problem

• EEG denoising
• unsupervised training
• wearable EEG
• noise artifacts
• self-supervision

Innovation

• self-supervised denoising
• intelligent partitioning
• wearable EEG
• deep learning
• zero-shot denoising