Federated Learning for Beginners: Types, Simulation Environments, and Open Challenges

Federated Learning (FL) is a ¹ decentralized machine learning (DML) technique. It was introduced by Google research team in 2017, where several clients can train a model collaboratively without sharing their raw data. Unlike centralized machine learning techniques, FL is able to preserve privacy. Since its inception, FL has gained widespread attention and has been the focus of extensive research, resulting in the development of various novel protocols, frameworks, and techniques to address its challenges. Despite these advancements, several research challenges in FL remain unexplored. While there are numerous research articles and tutorials available on FL, a beginner’s guide to this domain is lacking in the literature. Therefore, this paper targets to motivate novice researchers by presenting a simplified and systematic introduction to FL. In addition to introducing FL fundamentals, this paper also highlights some open research areas in the field.