Communication for Federated Learning
Title: Communication for Federated Learning
DNr: LiU-gpu-2022-1
Project Type: LiU Compute
Principal Investigator: Ema Becirovic <ema.becirovic@liu.se>
Affiliation: Linköpings universitet
Duration: 2022-02-16 – 2023-03-01
Classification: 20203
Keywords:

Abstract

In this work we study different communication transmission schemes for federated learning. The difference from conventional transmission schemes is that in federated learning the goal is not to necessarily achieve high data throughput, rather to properly train the targeted machine learning problems. Both analog and digital communication is studied.