Federated Machine Learning: Concept and Applications

TL;DR

The paper addresses the challenge of data islanding and privacy in AI by presenting federated learning as a secure, decentralized framework. It formalizes definitions, categorizes approaches into horizontal, vertical, and federated transfer learning, and outlines architectures and security notions for each. It surveys privacy-preserving techniques (SMC, differential privacy, homomorphic encryption), discusses indirect leakage risks, and explores incentive mechanisms and data alliances, including blockchain-based provenance. The work highlights applications across finance, retail, and healthcare, and envisions cross-organization data networks that preserve privacy while enabling knowledge sharing and collaboration.

Abstract

Today's AI still faces two major challenges. One is that in most industries, data exists in the form of isolated islands. The other is the strengthening of data privacy and security. We propose a possible solution to these challenges: secure federated learning. Beyond the federated learning framework first proposed by Google in 2016, we introduce a comprehensive secure federated learning framework, which includes horizontal federated learning, vertical federated learning and federated transfer learning. We provide definitions, architectures and applications for the federated learning framework, and provide a comprehensive survey of existing works on this subject. In addition, we propose building data networks among organizations based on federated mechanisms as an effective solution to allow knowledge to be shared without compromising user privacy.

Figures (5)

• Figure 1: GDPR: EU regulation on data protection
• Figure 2: Categorization of Federated Learning
• Figure 3: Architecture for a horizontal federated learning system
• Figure 4: Architecture for a vertical federated learning system
• Figure 5: Data Alliance Allocates the Benefits on Blockchain