FedHeN: Federated Learning in Heterogeneous Networks

Durmus Alp Emre Acar; Venkatesh Saligrama

FedHeN: Federated Learning in Heterogeneous Networks

Machine Learning 2022-07-08 v1 Distributed, Parallel, and Cluster Computing

Authors: Durmus Alp Emre Acar , Venkatesh Saligrama

Abstract

We propose a novel training recipe for federated learning with heterogeneous networks where each device can have different architectures. We introduce training with a side objective to the devices of higher complexities to jointly train different architectures in a federated setting. We empirically show that our approach improves the performance of different architectures and leads to high communication savings compared to the state-of-the-art methods.

Keywords

federated learning neural network training deep learning

Cite

@article{arxiv.2207.03031,
  title  = {FedHeN: Federated Learning in Heterogeneous Networks},
  author = {Durmus Alp Emre Acar and Venkatesh Saligrama},
  journal= {arXiv preprint arXiv:2207.03031},
  year   = {2022}
}

Comments

Workshop paper to be appear at DyNN, ICML 2022

Related papers

View all related →