FedRolex: Model-Heterogeneous Federated Learning with Rolling Sub-Model Extraction

Most cross-device federated learning (FL) studies focus on the model-homogeneous setting where the global server model and local client models are identical. However, such constraint not only excludes low-end clients who would otherwise make unique contributions to model training but also restrains...

Descripció completa

Dades bibliogràfiques

Autors principals:	Alam, Samiul, Liu, Luyang, Yan, Ming, Zhang, Mi
Format:	Article
Publicat:	2022
Matèries:	Machine Learning Cryptography and Security Computer Vision and Pattern Recognition Distributed, Parallel, and Cluster Computing
Accés en línia:	https://arxiv.org/abs/2212.01548 http://arxiv.org/abs/2212.01548
Etiquetes:	Afegir etiqueta Sense etiquetes, Sigues el primer a etiquetar aquest registre!