'Decentralized Personalized Federated Learning based on a Conditional Sparse-to-Sparser Scheme'
Published: 6 September 2023
Our Distributed AI paper 'Decentralized Personalized Federated Learning based on a Conditional Sparse-to-Sparser Scheme' authored by Qianyu Long, Qiyuan Wang, Christos Anagnostopoulos, and Daning Bi is now available on arXiv (arXiv:2404.15943). Keywords: Distributed AI, Dynamic aggregation, Personalized Federated Learning.
Link: arXiv/DOI
Link: Source code
We propose a novel sparse-to-sparser training scheme: DA-DPFL. DA-DPFL initializes with a subset of model parameters, which progressively reduces during training via dynamic aggregation and leads to substantial energy savings while retaining adequate information during critical learning periods.
First published: 6 September 2023
Link: arXiv/DOI
Link: Source code