Guardado en:
Detalles Bibliográficos
Autores principales: Lixandru, Andrei, van Gerven, Marcel, Pequito, Sergio
Formato: Preprint
Publicado: 2024
Materias:
Acceso en línea:https://arxiv.org/abs/2412.02546
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866915047015448576
author Lixandru, Andrei
van Gerven, Marcel
Pequito, Sergio
author_facet Lixandru, Andrei
van Gerven, Marcel
Pequito, Sergio
contents Distributed optimization is fundamental to modern machine learning applications like federated learning, but existing methods often struggle with ill-conditioned problems and face stability-versus-speed tradeoffs. We introduce fractional order distributed optimization (FrODO); a theoretically-grounded framework that incorporates fractional-order memory terms to enhance convergence properties in challenging optimization landscapes. Our approach achieves provable linear convergence for any strongly connected network. Through empirical validation, our results suggest that FrODO achieves up to 4 times faster convergence versus baselines on ill-conditioned problems and 2-3 times speedup in federated neural network training, while maintaining stability and theoretical guarantees.
format Preprint
id arxiv_https___arxiv_org_abs_2412_02546
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Fractional Order Distributed Optimization
Lixandru, Andrei
van Gerven, Marcel
Pequito, Sergio
Machine Learning
Distributed optimization is fundamental to modern machine learning applications like federated learning, but existing methods often struggle with ill-conditioned problems and face stability-versus-speed tradeoffs. We introduce fractional order distributed optimization (FrODO); a theoretically-grounded framework that incorporates fractional-order memory terms to enhance convergence properties in challenging optimization landscapes. Our approach achieves provable linear convergence for any strongly connected network. Through empirical validation, our results suggest that FrODO achieves up to 4 times faster convergence versus baselines on ill-conditioned problems and 2-3 times speedup in federated neural network training, while maintaining stability and theoretical guarantees.
title Fractional Order Distributed Optimization
topic Machine Learning
url https://arxiv.org/abs/2412.02546