Privacy-Preserving Federated Learning

This book offers a systematic, self-contained treatment of privacy-preserving federated learning (PPFL). It covers the complete arc from foundations to deployment. This book also illustrates how federated learning works and where it leaks; the privacy threats that matter in practice, from gradient inversion to membership inference; the protection techniques-differential privacy, secure multi-party computation, homomorphic encryption, and trusted execution environment. The authors provide the metrics and benchmarks needed to evaluate them and advanced topics including federated unlearning, vertical federated learning, and privacy-preserving fine-tuning of large language models.

Beyond algorithms, this book examines real-world systems and governance: open-source frameworks such as Flower, FATE, and NVIDIA FLARE; deployments in healthcare, finance, and smart cities. The regulatory landscape shaped by GDPR, HIPAA, CCPA, and PIPL. Framework comparisons, threat models, and case studies connect theory to engineering and policy throughout are included in this book as well.

Graduate students and researchers will find a rigorous entry point into the field when reading this book as well as engineers and system architects, concrete deployment guidance and policy professionals. Chapters are largely self-contained, supporting role-based reading paths and use as both course text and desk reference.