"Federated Learning: How Private Is It Really?"

Federated Learning (FL) is a popular structure that enables one to learn a Machine Learning (ML) model collaboratively. The classical FL structure involves multiple clients, each with their own local data that they may want to keep private, and a server responsible for learning a global ML model. One of the main reasons for FL's popularity is that clients can keep their data private while still benefiting from combined learning across all of their data. Saurabh Bagchi, a Purdue University professor of Electrical and Computer Engineering and Computer Science, and Arash Nourian, General Manager/Director of Engineering at AWS AI, discuss the ongoing back and forth over protecting data privacy through FL. This article continues to discuss experts' insights on the concept of FL and potential data leakage attacks. 

CACM reports "Federated Learning: How Private Is It Really?"