Abstract
In Federated Learning (FL), clients with low computational power train a common machine model by exchanging parameters via updates instead of transmitting potentially private data. Federated Dropout (FD) is a technique that improves the communication efficiency of a FL session by selecting a subset of model parameters to be updated in each training round. However, compared to standard FL, FD produces considerably lower accuracy and faces a longer convergence time. In this chapter, we leverage coding theory to enhance FD by allowing different sub-models to be used at each client. We also show that by carefully tuning the server learning rate hyper-parameter, we can achieve higher training speed while also reaching up to the same final accuracy as the no dropout case. Evaluations on the EMNIST dataset show that our mechanism achieves 99.6% of the final accuracy of the no dropout case while requiring 2.43 × less bandwidth to achieve this level of accuracy.
| Original language | English (US) |
|---|---|
| Title of host publication | Trustworthy Federated Learning |
| Publisher | Springer International Publishing |
| Pages | 84-99 |
| Number of pages | 16 |
| ISBN (Print) | 9783031289958 |
| DOIs | |
| State | Published - Mar 29 2023 |
| Externally published | Yes |
Bibliographical note
KAUST Repository Item: Exported on 2023-05-24Acknowledged KAUST grant number(s): ORA-CRG2021-4699
Acknowledgements: This research work was conducted with funding awarded by the Swedish Research Council for the project “Scalable Federated Learning” with registration number 2021-04610. This publication is based upon work supported by the King Abdullah University of Science and Technology (KAUST) Office of Research Administration (ORA) under Award No. ORA-CRG2021-4699.
This publication acknowledges KAUST support, but has no KAUST affiliated authors.