Mosaic Learning: A Framework for Decentralized Learning with Model Fragmentation

Decentralized learning (DL) enables collaborative machine learning (ML) without a central server, making it suitable for settings where training data cannot be centrally hosted. We introduce Mosaic Learning, a DL framework that decomposes models into fragments and disseminates them independently acr...

Descripción completa

Detalles Bibliográficos

Autores principales:	Biswas, Sayan, Frey, Davide, Gaudel, Romaric, Gupta, Nirupam, Kermarrec, Anne-Marie, Lerévérend, Dimitri, Pires, Rafael, Sharma, Rishi, Taïani, François, de Vos, Martijn
Formato:	Artículo
Publicado:	2026
Materias:	Machine Learning
Acceso en línea:	https://arxiv.org/abs/2602.04352 http://arxiv.org/abs/2602.04352
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!