Federated Learning: New Approach to Building AI Models

Leveraging Open Source Tools to Accelerate Technology Development across Organizations and Regions

Traditional machine learning (ML) models are centralized and involve vast amounts of data. However, both the urgency to guarantee data privacy and to abide by strict regulations imposed across regions have contributed to the emergence of a new and powerful alternative technique, federated learning. Instead of acquiring data from a central server or cloud, federated learning allows localized model training. The technique ensures privacy preservation, and better global models are trained without exchanging raw data that holds private and sensitive information. Attracted to this powerful privacy-protecting technique, a growing number of market participants, academics, and end-use industries are adopting federated learning at an unprecedented rate.

Federated learning is a distributed ML architecture that enables a global model to be trained using decentralized data. It is intended to utilize data from across an organization accurately and effectively. To help companies gain valuable insights about this emerging technique, this report offers an overview of the federated learning industry, market dynamics, key market players, research directions, key application areas, and recent developments.

The following chapters are included:

• Overview of federated learning
• Market forecast, drivers, and challenges
• Key research directions for federated learning
• IP landscape analysis
• Key enablers and recent technology developments
• Companies to action, including Edgify, Owkin, Fetch.ai, Sherpa Europe, and WeBank
• Growth opportunities