Seminar on April 29, 2022

Title

Cross-device federated learning and federated analytics at Google

Speaker

When and Where

April 29, 2022, 15:30-16:30
Online, see registration link

Abstract

Federated learning enables mobile devices to collaboratively train a machine learning model while keeping raw data local on each device. Federated analytics enables collaboratively computing statistical aggregates for data science queries. Google’s system for federated learning and analytics that runs on hundreds of millions of Android phones and combines techniques from distributed systems, machine learning, differential privacy, secure multiparty computation, and more. This talk will give an overview of how Google’s system works and some of the key results that enable its use in improving privacy and utility in Google apps and services.

Short Bio

Daniel Ramage has been a research scientist at Google in Mountain View since 2011. He got his PhD from Stanford. His undergraduate degrees from MIT are in Computer Science and Math. Prior to Stanford, he spent some time at IBM’s Zurich Research Lab working on data mining. In Fall 2009, he worked at Microsoft Research in Redmond. His research interests include machine learning, mobile systems, natural language processing, and security/privacy.