计算机科学
人工智能
人工神经网络
睡眠(系统调用)
模式识别(心理学)
特征(语言学)
语音识别
尖峰神经网络
睡眠阶段
光学(聚焦)
钥匙(锁)
作者
Xinzhong Cui,Xi Wu,Xiaoli Li
出处
期刊:
[Institute of Electrical and Electronics Engineers]
日期:2026-04-21
卷期号:: 8332-8336
标识
DOI:10.1109/icassp55912.2026.11462764
摘要
Sleep is vital for human health, and sleep staging is a key technology for evaluating sleep quality. Rapid progress in deep learning has led to significant advances in automatic sleep staging. However, designing lightweight algorithms suitable for low-power wearable devices remains challenging. This study proposes a lightweight spiking neural network based on attention mechanisms, named LASpikeSleep. The model leverages the temporal modeling capability of spiking neural networks by segmenting electroencephalogram (EEG) signals into time steps, followed by temporal attention to capture the dependencies among segmented signals. In parallel, channel attention embedded within spiking neurons optimizes membrane potentials. These mechanisms regulate spike firing, enabling lightweight and low-power automatic sleep staging. Experimental results demonstrate that the LASpikeSleep delivers competitive performance with superior energy efficiency, underscoring its potential for wearable sleep monitoring applications.
科研通智能强力驱动
Strongly Powered by AbleSci AI