Sound-Based Sleep Staging Using Semi-Supervised Learning
Dr. Dong Li, Department of Computer Science and Electrical Engineering.
This project develops machine learning methods for classifying sleep stages from audio recordings, offering a non-invasive alternative to clinical polysomnography. Using the PSG-Audio dataset, we implement a transformer-based model that processes Mel spectrograms of nocturnal breathing sounds to predict four sleep stages.
Sleep Disorder Monitoring
Dr. Dong Li, Department of Computer Science and Electrical Engineering.
This project aims to create a sleep stage monitoring system that utilizes contact-free multi-modal data. The long term goal is to extend the system to be able to detect and monitor sleep disorders. By using cutting-edge signal processing and machine learning methods, the system will be capable of providing precise and non-invasive sleep pattern analysis. In order to make the system scalable and capable of real-time execution, the ultimate solution will be refined for deployment on edge devices.