Multi-Level Branched Regularization for Federated Learning

Jinkyu Kim, Geeho Kim, Bohyung Han
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:11058-11073, 2022.

Abstract

A critical challenge of federated learning is data heterogeneity and imbalance across clients, which leads to inconsistency between local networks and unstable convergence of global models. To alleviate the limitations, we propose a novel architectural regularization technique that constructs multiple auxiliary branches in each local model by grafting local and global subnetworks at several different levels and that learns the representations of the main pathway in the local model congruent to the auxiliary hybrid pathways via online knowledge distillation. The proposed technique is effective to robustify the global model even in the non-iid setting and is applicable to various federated learning frameworks conveniently without incurring extra communication costs. We perform comprehensive empirical studies and demonstrate remarkable performance gains in terms of accuracy and efficiency compared to existing methods. The source code is available at our project page.

Cite this Paper


BibTeX
@InProceedings{pmlr-v162-kim22a, title = {Multi-Level Branched Regularization for Federated Learning}, author = {Kim, Jinkyu and Kim, Geeho and Han, Bohyung}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {11058--11073}, year = {2022}, editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan}, volume = {162}, series = {Proceedings of Machine Learning Research}, month = {17--23 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v162/kim22a/kim22a.pdf}, url = {https://proceedings.mlr.press/v162/kim22a.html}, abstract = {A critical challenge of federated learning is data heterogeneity and imbalance across clients, which leads to inconsistency between local networks and unstable convergence of global models. To alleviate the limitations, we propose a novel architectural regularization technique that constructs multiple auxiliary branches in each local model by grafting local and global subnetworks at several different levels and that learns the representations of the main pathway in the local model congruent to the auxiliary hybrid pathways via online knowledge distillation. The proposed technique is effective to robustify the global model even in the non-iid setting and is applicable to various federated learning frameworks conveniently without incurring extra communication costs. We perform comprehensive empirical studies and demonstrate remarkable performance gains in terms of accuracy and efficiency compared to existing methods. The source code is available at our project page.} }
Endnote
%0 Conference Paper %T Multi-Level Branched Regularization for Federated Learning %A Jinkyu Kim %A Geeho Kim %A Bohyung Han %B Proceedings of the 39th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2022 %E Kamalika Chaudhuri %E Stefanie Jegelka %E Le Song %E Csaba Szepesvari %E Gang Niu %E Sivan Sabato %F pmlr-v162-kim22a %I PMLR %P 11058--11073 %U https://proceedings.mlr.press/v162/kim22a.html %V 162 %X A critical challenge of federated learning is data heterogeneity and imbalance across clients, which leads to inconsistency between local networks and unstable convergence of global models. To alleviate the limitations, we propose a novel architectural regularization technique that constructs multiple auxiliary branches in each local model by grafting local and global subnetworks at several different levels and that learns the representations of the main pathway in the local model congruent to the auxiliary hybrid pathways via online knowledge distillation. The proposed technique is effective to robustify the global model even in the non-iid setting and is applicable to various federated learning frameworks conveniently without incurring extra communication costs. We perform comprehensive empirical studies and demonstrate remarkable performance gains in terms of accuracy and efficiency compared to existing methods. The source code is available at our project page.
APA
Kim, J., Kim, G. & Han, B.. (2022). Multi-Level Branched Regularization for Federated Learning. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:11058-11073 Available from https://proceedings.mlr.press/v162/kim22a.html.

Related Material