Open Source Summit North America 2023

Federated learning is a collaborative machine learning technique that allows multiple clients to train an algorithm without sharing raw data. This technique enables enterprises to collaborate, break data silos, and mine the value of discrete data distributed across different organizations while complying with relevant privacy protection laws and regulations. The FATE and OpenFL frameworks are the two widely used open-source industrial-grade federated learning frameworks that have been applied in various industries, such as finance and healthcare, to solve complex problems. However, the adoption of federated learning across organizations raises new challenges in orchestration and lifecycle management. In this talk, we will introduce the concept, principles, and typical use-cases of federated learning, followed by an introduction to the FATE and OpenFL frameworks. We will then discuss the challenges of orchestrating federated learning in a production environment and introduce the FedLCM (Federation Lifecycle Manager) project, a unified experience for managing different federated learning frameworks in a multi-cloud environment. This talk will be beneficial for those who are interested in federated learning and want to learn about the available frameworks and their orchestration challenges.