Federated Learning and Analytics at Google - Workshop Introduction

Explore federated learning and analytics at Google in this 31-minute TechTalk presented by Sean Augenstein and Brendan McMahan. Dive into the concept of decentralized and privacy-preserving machine learning, pioneered by McMahan's team. Learn about the four privacy principles, cross-device federated learning, data minimization techniques, and the tradeoffs involved. Discover practical applications like Gboard privacy and neural network training. Gain insights into Tensorflow Federated, federated analytics, and cutting-edge research topics such as federated dropout and embedding selection. Engage with the speakers' expertise in online learning, large-scale convex optimization, and reinforcement learning as they address questions and discuss the future of privacy-preserving machine learning techniques.

Intro
Welcome
Sean Augustine
Zach Charles
Goals
Mini Workshops
Brendan McMahon
Four Privacy Principles
Crossdevice FL
Data minimization
Observations
Tradeoffs
Neural Network Training
Gboard Privacy
DPFTRL
Tensorflow Federated
Federated Analytics
Research Question
Federated Dropout
Embedding Selection
Questions