Authored by researchers and practitioners who build cutting-edge federated learning applications to solve real-world problems, this book covers the spectrum of federated learning technology from concepts and application scenarios to advanced algorithms and finally system implementation in three parts. It provides a comprehensive review and summary of federated learning technology, as well as presenting numerous novel federated learning algorithms which no other books have summarized. The work also references the most recent papers, articles and reviews from the past several years to keep pace with the academic and industrial state of the art of federated learning.

The first part lays a foundational understanding of federated learning by going through its definition and characteristics, and also possible application scenarios and related privacy protection technologies. The second part elaborates on some of the federated learning algorithms innovated by JD Technology which encompass both vertical and horizontal scenarios, including vertical federated tree models, linear regression, kernel learning, asynchronous methods, deep learning, homomorphic encryption, and reinforcement learning. The third and final part shifts in scope to federated learning systems — namely JD Technology's own FedLearn system — by discussing its design and implementation using gRPC, in addition to specific performance optimization techniques plus integration with blockchain technology.

This book will serve as a great reference for readers who are experienced in federated learning algorithms, building privacy-preserving machine learning applications or solving real-world problems with privacy-restricted scenarios.

Contents:

• 
  

  Federated Learning Knowledge:

  
  
  
  
  • Introduction to Federated Learning
  
  
  • Federated Learning Application Scenarios
  
  
  • Common Privacy Protection Technologies
  
  
  
  


• 
  

  Federated Learning Algorithms:

  
  
  
  
  • Tree-Based Models in Vertical Federated Learning
  
  
  • Vertical Federated Linear Regression Algorithm
  
  
  • Vertically Federated Kernel Learning
  
  
  • Asynchronous Vertical Federated Learning Algorithm
  
  
  • Secure Bilevel Asynchronous Vertical Federated Learning with Backward Updating
  
  
  • Vertical Federated Deep Learning Algorithms
  
  
  • Faster Secure Data Mining Framework via Homomorphic Encryption
  
  
  • Horizontal Federated Learning Algorithms
  
  
  • Mixed Federated Learning Algorithms
  
  
  • Federated Reinforcement Learning
  
  
  
  


• 
  

  Federated Learning Systems:

  
  
  
  
  • Detailed Exploration of the FedLearn Federated Learning System
  
  
  • Application of gRPC in FedLearn
  
  
  • Performance Optimization Practices in Real-World Scenarios
  
  
  • Federated Learning Based on Blockchain
  
  
  
  

Readership: Advanced undergraduate and graduate students, researchers and practitioners with somewhat knowledge about machine learning, distributed system and privacy preserving technologies. This book will serve as a greate reference for readers who has experiences of federated learning algorithms, building privacy preserving machine learning applications or solving real-world problems with privacy-restricted scenarios.

Key Features:

• Presents numerous novel federated learning algorithms which no other books have summarized


• References the most recent papers, articles and reviews from the past several years to keeping pace with the academic and industry state of the art of federated learning


• Authors are researchers and practitioners who build cutting-edge federated learning applications to solve real-world problems; such experience at the forefront of federated learning is unique