Federated Learning: Privacy by Design for Machine Learning

18 May 2021
by Arushi Pandey, Nisha Krishan

Enterprises have identified Artificial Intelligence (AI) as a quintessential enabling layer in their digital transformations. As AI adoption increases, so does the concern for user data privacy, coupled with the need for an agency to safeguard it. Recently, government guidelines, regulations, and bills such as the GDPR and the CCPA have been introduced to improve data privacy and governance. These guidelines have explicit rules regarding the storage, processing, and usage of data for any purpose, and any breach is heavily penalized.

This is where the concept of federated learning comes into the picture. Federated learning is a method of training machine-learning models in a way that user data does not leave its location, and hence remains safe and private. With its privacy-by-design methodology, federated learning presents immense potential and use across data-critical industries and use cases.


All industries and geographies


In this report we study the following themes:

  • Define federated learning and its key characteristics
  • Compare federated learning and centralized learning
  • Detail the benefits and application areas of federated learning
  • Provide an overview of the ecosystem players for federated learning
  • Introduce a decision-making framework for federated learning adoption


Digital Services

Sourcing and Vendor Management


Page Count: 13