Layer-wise Adaptive Model Aggregation for Scalable Federated Learning
AAAI, 2023
This study proposes a communication-efficient model aggregation method for federated learning. The layer-wise aggregation frequency is adjusted based on online gradient magnitude.
BibTeX
@inproceedings{lee2023layer,
title={Layer-wise adaptive model aggregation for scalable federated learning},
author={Lee, Sunwoo and Zhang, Tuo and Avestimehr, A Salman},
booktitle={Proceedings of the AAAI conference on artificial intelligence},
volume={37},
number={7},
pages={8491--8499},
year={2023}
}