M31.8 CONNECT THE MECHANISM
Coordinate learning while data remain distributed
Your phone's keyboard gets better at predicting your next word, yet your messages never leave the phone. See how millions of phones train one model, and what privacy that does and doesn't buy.
LESSON OVERVIEW15 min lesson
Lesson overview
Your phone's keyboard gets better at predicting your next word, yet your messages never leave the phone. See how millions of phones train one model, and what privacy that does and doesn't buy.
What you’ll explore
- Compute a FedAvg round as a data-weighted average, explain how non-IID clients and local steps cause drift, weigh communication cost, and separate what secure aggregation and differential privacy each protect.
GO TO THE SOURCE
Original explanations, connected to the research.
Communication-Efficient Learning of Deep Networks from Decentralized Data (McMahan et al., 2017), the FedAvg paperFederated Learning for Mobile Keyboard Prediction (Hard et al., 2018)Practical Secure Aggregation for Privacy-Preserving Machine Learning (Bonawitz et al., 2017)Federated Learning with Formal Differential Privacy Guarantees (Google Research blog, 2022)Federated Learning of Gboard Language Models with Differential Privacy (Xu et al., 2023)Suggest a correction
A precise note can make an explanation better.
Choose the scene and describe what needs attention. Download a feedback file to share through a channel you already use. This page does not send feedback or connect you with a reviewer.