SECURITY ISSUES OF FEDERATED LEARNING IN REAL-LIFE APPLICATIONS

Abstract

Machine Learning (ML) is becoming one of the most popular and widely used IT technologies in the past 10 years. The sharing and analysing of large volumes of data promises to revolutionalise many sectors, such as transport, healthcare and defence. This data's value and the consequent competitive advantages from its processing have attracted significant adversarial efforts against its security, privacy and availability. Recent advancements in federated learning (FL) show promising results in protecting data security and privacy and equally create additional opportunities for organised cyber criminals to capitalise from its use. This paper presents the existing and emerging security threats against FL using real-life scenarios and applications.