FLUTE: Breaking Barriers for Federated Learning Research at Scale

Опубликовано 19 мая 2022, 14:04

In this video we introduce FLUTE: Federated Learning Utilities and Tools for Experimentation, a framework for running large-scale offline FL simulations. FLUTE platform enables researchers and developers to investigate proof-of-concept implementations and validate their performance, all designed to lower the entry barriers to apply Federated Learning in their ML solutions. FLUTE's architecture is designed to be scalable and flexible allowing quick prototyping of novel approaches in federation, optimization, quantization, privacy and more.

GitHub: aka.ms/msrflute
Blog post: aka.ms/msrflute-blog
Project FLUTE: microsoft.com/en-us/research/p...