Facial Image Recognition in Distributed Machine Learning Using Rich Clients

Abstract

Most of current research into edge computing involves using devices only for sensing and communicating data. In contrast, by offloading machine learning on the device that constitutes the source of the data, one can construct a system with better privacy protection and reduced communication cost. We therefore consider an edge computing model in which rich clients also participate in machine learning. Experiments conducted using facial images confirmed the benefits of the proposed model in terms of privacy and communication costs.