Distributed autoencoder classifier network for small‐scale and scattered <scp>COVID</scp>‐19 dataset classification

Abstract In healthcare, small‐scare data are stored with individual entities, such as hospitals, and they not shared. However, one entity sufficient for training a machine learning model therefore cannot be fully utilized. Given that large amount of small‐scale is widely distributed between hospitals/individuals, it necessary to deploy an easy, scalable, secure computational framework. We aim aggregate these scattered train neural networks achieve classification detection on coronavirus disease 2019 (COVID‐19) datasets. propose autoencoder (AE) classifier network this purpose. It contains central multiple AEs. The AEs used generators. A local generator uses actual COVID‐19 computed tomography image the input outputs synthetic image. well‐trained provides model. learns information from all generated using AE. Experiments performed some AE outperforms models use single subset, its performance similar regular classifier. proposed solves problem while ensuring privacy. accuracy same achieved entire data.