Implementing ultra-lightweight co-inference model in ubiquitous edge device for atrial fibrillation detection

Jiarong Chen, Mingzhe Jiang, Xianbin Zhang, Daniel Santos da Silva, Victor Hugo C. de Albuquerque, Wanqing Wu

Abstract: Highlights•A framework is proposed for extremely resource-constrained edge devices.•Neural network can be ultra-lightweight with negligible performance loss.•The model training process can be done effectively without cloud participation.•A low-cost and real-time solution proposed for atrial fibrillation detection.