Towards Multi-Source Domain Generalization for Sleep Staging with Noisy Labels

TL;DR

This paper introduces a new benchmark and a robust multimodal sleep staging framework, FF-TRUST, that effectively handles domain shifts and noisy labels across diverse datasets.

Contribution

It presents the first benchmark for noisy labels in multi-source domain-generalized sleep staging and proposes FF-TRUST, a novel framework with joint regularization for improved robustness.

Findings

FF-TRUST achieves state-of-the-art results across multiple datasets.

Existing noisy-label methods perform poorly under domain shifts.

The benchmark facilitates future research in noisy, multi-source sleep staging.

Abstract

Automatic sleep staging is a multimodal learning problem involving heterogeneous physiological signals such as EEG and EOG, which often suffer from domain shifts across institutions, devices, and populations. In practice, these data are also affected by noisy annotations, yet label-noise-robust multi-source domain generalization remains underexplored. We present the first benchmark for Noisy Labels in Multi-Source Domain-Generalized Sleep Staging (NL-DGSS) and show that existing noisy-label learning methods degrade substantially when domain shifts and label noise coexist. To address this challenge, we propose FF-TRUST, a domain-invariant multimodal sleep staging framework with Joint Time-Frequency Early Learning Regularization (JTF-ELR). By jointly exploiting temporal and spectral consistency together with confidence-diversity regularization, FF-TRUST improves robustness under noisy supervision. Experiments on five public datasets demonstrate consistent state-of-the-art performance under diverse symmetric and asymmetric noise settings. The benchmark and code will be made publicly available at https://github.com/KNWang970918/FF-TRUST.git.