Splitfed learning processing for IoT and Big Data applications

Federated learning (FL) and split learning (SL) are two approaches to learning that can greatly enhance ubiquitous intelligence within the Internet of Things (IoT). FL involves combining trained machine learning (ML) models, each using data into a global model. On the one hand SL allows different segments of an ML model to be collaboratively trained on workers in a learning framework. While both FL and SL have their strengths and limitations, they can work together to achieve intelligence in the IoT. Recently, there has been a surge of interest in exploring the synergy between FL and SL making it a key focus area for research. This chapter offers insights into the advancements in FL and SL and provides an overview of cutting-edge technologies for integrating these approaches into an edge computing-based IoT setting. Furthermore, it addresses some existing challenges. The chapter also discusses future research directions to spark further exploration, within the academic community.