FedScale: Benchmarking Model and System Performance of Federated Learning at Scale

Fan Lai, Yinwei Dai, Sanjay Singapuram, Jiachen Liu, Xiangfeng Zhu, Harsha Madhyastha, Mosharaf Chowdhury
Proceedings of the 39th International Conference on Machine Learning, PMLR 162:11814-11827, 2022.

Abstract

We present FedScale, a federated learning (FL) benchmarking suite with realistic datasets and a scalable runtime to enable reproducible FL research. FedScale datasets encompass a wide range of critical FL tasks, ranging from image classification and object detection to language modeling and speech recognition. Each dataset comes with a unified evaluation protocol using real-world data splits and evaluation metrics. To reproduce realistic FL behavior, FedScale contains a scalable and extensible runtime. It provides high-level APIs to implement FL algorithms, deploy them at scale across diverse hardware and software backends, and evaluate them at scale, all with minimal developer efforts. We combine the two to perform systematic benchmarking experiments and highlight potential opportunities for heterogeneity-aware co-optimizations in FL. FedScale is open-source and actively maintained by contributors from different institutions at http://fedscale.ai. We welcome feedback and contributions from the community.

Cite this Paper

BibTeX
@InProceedings{pmlr-v162-lai22a, title = {{F}ed{S}cale: Benchmarking Model and System Performance of Federated Learning at Scale}, author = {Lai, Fan and Dai, Yinwei and Singapuram, Sanjay and Liu, Jiachen and Zhu, Xiangfeng and Madhyastha, Harsha and Chowdhury, Mosharaf}, booktitle = {Proceedings of the 39th International Conference on Machine Learning}, pages = {11814--11827}, 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/lai22a/lai22a.pdf}, url = {https://proceedings.mlr.press/v162/lai22a.html}, abstract = {We present FedScale, a federated learning (FL) benchmarking suite with realistic datasets and a scalable runtime to enable reproducible FL research. FedScale datasets encompass a wide range of critical FL tasks, ranging from image classification and object detection to language modeling and speech recognition. Each dataset comes with a unified evaluation protocol using real-world data splits and evaluation metrics. To reproduce realistic FL behavior, FedScale contains a scalable and extensible runtime. It provides high-level APIs to implement FL algorithms, deploy them at scale across diverse hardware and software backends, and evaluate them at scale, all with minimal developer efforts. We combine the two to perform systematic benchmarking experiments and highlight potential opportunities for heterogeneity-aware co-optimizations in FL. FedScale is open-source and actively maintained by contributors from different institutions at http://fedscale.ai. We welcome feedback and contributions from the community.} }
Endnote
%0 Conference Paper %T FedScale: Benchmarking Model and System Performance of Federated Learning at Scale %A Fan Lai %A Yinwei Dai %A Sanjay Singapuram %A Jiachen Liu %A Xiangfeng Zhu %A Harsha Madhyastha %A Mosharaf Chowdhury %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-lai22a %I PMLR %P 11814--11827 %U https://proceedings.mlr.press/v162/lai22a.html %V 162 %X We present FedScale, a federated learning (FL) benchmarking suite with realistic datasets and a scalable runtime to enable reproducible FL research. FedScale datasets encompass a wide range of critical FL tasks, ranging from image classification and object detection to language modeling and speech recognition. Each dataset comes with a unified evaluation protocol using real-world data splits and evaluation metrics. To reproduce realistic FL behavior, FedScale contains a scalable and extensible runtime. It provides high-level APIs to implement FL algorithms, deploy them at scale across diverse hardware and software backends, and evaluate them at scale, all with minimal developer efforts. We combine the two to perform systematic benchmarking experiments and highlight potential opportunities for heterogeneity-aware co-optimizations in FL. FedScale is open-source and actively maintained by contributors from different institutions at http://fedscale.ai. We welcome feedback and contributions from the community.
APA
Lai, F., Dai, Y., Singapuram, S., Liu, J., Zhu, X., Madhyastha, H. & Chowdhury, M.. (2022). FedScale: Benchmarking Model and System Performance of Federated Learning at Scale. Proceedings of the 39th International Conference on Machine Learning, in Proceedings of Machine Learning Research 162:11814-11827 Available from https://proceedings.mlr.press/v162/lai22a.html.

Related Material