A Review of Split Learning and Federated Learning: Challenges and Synergies

Abstract

Split Learning and Federated Learning have emerged as key techniques in the domain of privacy-preserving distributed machine learning. This paper reviews the recent developments in both paradigms, discussing their respective advantages, limitations, and the potential for their integration. We provide an analysis of current research trends, explore challenges in implementation, and suggest future directions for improving these approaches. The review serves as a resource for researchers and practitioners interested in the evolving landscape of distributed machine learning.