Tackling the Complexities of Federated Learning
Raj Thakur, Shreyansh Patel, Neelesh Singh, Aaryan Barde, Snehlata Barde · 2025
Federated learning is an innovative machine learning approach enabling multiple devices to collaboratively train a model without sharing sensitive data, overseen by a central server. This chapter explores the strengths and vulnerabilities of federated learning, emphasizing its secure nature while acknowledging potential risks. It discusses key challenges like communication overhead and data heterogeneity, along with techniques to address them. The chapter also provides an overview of current federated learning methods, highlighting their efficacy and areas for improvement. Future directions are outlined, including enhancing robustness against attacks and scaling for larger systems, making this chapter a valuable resource for researchers and practitioners in the field of privacy-preserving machine learning.