Guardado en:
Detalles Bibliográficos
Autores principales: Yoon, TaeHo, Choudhury, Sayantan, Loizou, Nicolas
Formato: Preprint
Publicado: 2025
Materias:
Acceso en línea:https://arxiv.org/abs/2501.08263
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866917065979330560
author Yoon, TaeHo
Choudhury, Sayantan
Loizou, Nicolas
author_facet Yoon, TaeHo
Choudhury, Sayantan
Loizou, Nicolas
contents Traditional Federated Learning (FL) approaches assume collaborative clients with aligned objectives working towards a shared global model. However, in many real-world scenarios, clients act as rational players with individual objectives and strategic behaviors, a concept that existing FL frameworks are not equipped to adequately address. To bridge this gap, we introduce Multiplayer Federated Learning (MpFL), a novel framework that models the clients in the FL environment as players in a game-theoretic context, aiming to reach an equilibrium. In this scenario, each player tries to optimize their own utility function, which may not align with the collective goal. Within MpFL, we propose Per-Player Local Stochastic Gradient Descent (PEARL-SGD), an algorithm in which each player/client performs local updates independently and periodically communicates with other players. We theoretically analyze PEARL-SGD and prove that it reaches a neighborhood of equilibrium with less communication in the stochastic setup compared to its non-local counterpart. Finally, we verify our theoretical findings through numerical experiments.
format Preprint
id arxiv_https___arxiv_org_abs_2501_08263
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Multiplayer Federated Learning: Reaching Equilibrium with Less Communication
Yoon, TaeHo
Choudhury, Sayantan
Loizou, Nicolas
Machine Learning
Optimization and Control
Traditional Federated Learning (FL) approaches assume collaborative clients with aligned objectives working towards a shared global model. However, in many real-world scenarios, clients act as rational players with individual objectives and strategic behaviors, a concept that existing FL frameworks are not equipped to adequately address. To bridge this gap, we introduce Multiplayer Federated Learning (MpFL), a novel framework that models the clients in the FL environment as players in a game-theoretic context, aiming to reach an equilibrium. In this scenario, each player tries to optimize their own utility function, which may not align with the collective goal. Within MpFL, we propose Per-Player Local Stochastic Gradient Descent (PEARL-SGD), an algorithm in which each player/client performs local updates independently and periodically communicates with other players. We theoretically analyze PEARL-SGD and prove that it reaches a neighborhood of equilibrium with less communication in the stochastic setup compared to its non-local counterpart. Finally, we verify our theoretical findings through numerical experiments.
title Multiplayer Federated Learning: Reaching Equilibrium with Less Communication
topic Machine Learning
Optimization and Control
url https://arxiv.org/abs/2501.08263