EDEN: Communication-Efficient and Robust Distributed Mean Estimation for Federated Learning

Shay Vargaftik, Ran Ben Basat, Amit Portnoy, Gal Mendelson, Yaniv Ben Itzhak, Michael Mitzenmacher
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:21984-22014, 2022.

Abstract

Distributed Mean Estimation (DME) is a central building block in federated learning, where clients send local gradients to a parameter server for averaging and updating the model. Due to communication constraints, clients often use lossy compression techniques to compress the gradients, resulting in estimation inaccuracies. DME is more challenging when clients have diverse network conditions, such as constrained communication budgets and packet losses. In such settings, DME techniques often incur a significant increase in the estimation error leading to degraded learning performance. In this work, we propose a robust DME technique named EDEN that naturally handles heterogeneous communication budgets and packet losses. We derive appealing theoretical guarantees for EDEN and evaluate it empirically. Our results demonstrate that EDEN consistently improves over state-of-the-art DME techniques.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-vargaftik22a, title = {{EDEN}: Communication-Efficient and Robust Distributed Mean Estimation for Federated Learning}, author = {Vargaftik, Shay and Basat, Ran Ben and Portnoy, Amit and Mendelson, Gal and Itzhak, Yaniv Ben and Mitzenmacher, Michael}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {21984--22014}, 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/vargaftik22a/vargaftik22a.pdf}, url = {https://proceedings.mlr.press/v162/vargaftik22a.html}, abstract = {Distributed Mean Estimation (DME) is a central building block in federated learning, where clients send local gradients to a parameter server for averaging and updating the model. Due to communication constraints, clients often use lossy compression techniques to compress the gradients, resulting in estimation inaccuracies. DME is more challenging when clients have diverse network conditions, such as constrained communication budgets and packet losses. In such settings, DME techniques often incur a significant increase in the estimation error leading to degraded learning performance. In this work, we propose a robust DME technique named EDEN that naturally handles heterogeneous communication budgets and packet losses. We derive appealing theoretical guarantees for EDEN and evaluate it empirically. Our results demonstrate that EDEN consistently improves over state-of-the-art DME techniques.} }
Endnote
%0 Conference Paper %T EDEN: Communication-Efficient and Robust Distributed Mean Estimation for Federated Learning %A Shay Vargaftik %A Ran Ben Basat %A Amit Portnoy %A Gal Mendelson %A Yaniv Ben Itzhak %A Michael Mitzenmacher %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-vargaftik22a %I PMLR %P 21984--22014 %U https://proceedings.mlr.press/v162/vargaftik22a.html %V 162 %X Distributed Mean Estimation (DME) is a central building block in federated learning, where clients send local gradients to a parameter server for averaging and updating the model. Due to communication constraints, clients often use lossy compression techniques to compress the gradients, resulting in estimation inaccuracies. DME is more challenging when clients have diverse network conditions, such as constrained communication budgets and packet losses. In such settings, DME techniques often incur a significant increase in the estimation error leading to degraded learning performance. In this work, we propose a robust DME technique named EDEN that naturally handles heterogeneous communication budgets and packet losses. We derive appealing theoretical guarantees for EDEN and evaluate it empirically. Our results demonstrate that EDEN consistently improves over state-of-the-art DME techniques.
APA
Vargaftik, S., Basat, R.B., Portnoy, A., Mendelson, G., Itzhak, Y.B. & Mitzenmacher, M.. (2022). EDEN: Communication-Efficient and Robust Distributed Mean Estimation for Federated Learning. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:21984-22014 Available from https://proceedings.mlr.press/v162/vargaftik22a.html.

Related Material