Lightweight ML-Based Automatic Sleep Staging Framework with Constrained CNN and Mamba for Small-Sample EEG Datasets

2026-07-06

AI Summary

This study addresses the challenges of large model size, overfitting, deployment difficulty, and low recognition accuracy for the N1 and REM sleep stages in few-shot, single-channel EEG sleep staging. To this end, the authors propose GamSleepNet, a lightweight framework that integrates an enhanced Gabor convolution with learnable filters to construct a feature extraction block (FEB) and leverages a Mamba-based architecture for efficient temporal modeling. A novel contrastive loss function and a two-stage training strategy are also introduced. With only 30.86K parameters, the model achieves an overall accuracy of 87.86% on the Sleep-EDF dataset, significantly improving performance on challenging sleep stages and attaining state-of-the-art results across multiple metrics.

Abstract

Automatic sleep staging is a key technology for precise diagnosis and treatment of sleep disorders as well as long-term home sleep monitoring. Portable electroencephalogram (EEG) devices have become the focus of research due to their convenience in data collection. However, current methods still face three major challenges: large parameter sizes that easily lead to overfitting on small datasets, low accuracy in classifying difficult stages such as N1 and REM, unclear optimal training dataset size, and difficulty in deployment. This paper proposes GamSleepNet, a lightweight and low-latency automatic sleep staging framework for single-channel EEG. The framework features the FEB module, which combines improved Gabor kernels with learnable filters for feature extraction, uses the Mamba architecture to build a temporal classification network, introduces a novel contrastive loss and a two-stage training strategy, and experimentally validates the optimal dataset size for single-channel EEG sleep staging models. On the Sleepedf dataset, this model achieves an overall accuracy of 87.86 percent with only 30.86 thousand parameters, with all metrics reaching SOTA levels and significantly improving the identification accuracy of challenging sleep stages.

Problem

• automatic sleep staging
• small-sample EEG
• lightweight model
• sleep stage classification
• overfitting

Innovation

• lightweight sleep staging
• Mamba architecture
• Gabor-based feature extraction
• contrastive loss
• small-sample EEG