We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: FedGiA: An Efficient Hybrid Algorithm for Federated Learning

Abstract: Federated learning has shown its advances recently but is still facing many challenges, such as how algorithms save communication resources and reduce computational costs, and whether they converge. To address these critical issues, we propose a hybrid federated learning algorithm (FedGiA) that combines the gradient descent and the inexact alternating direction method of multipliers. The proposed algorithm is more communication- and computation-efficient than several state-of-the-art algorithms theoretically and numerically. Moreover, it also converges globally under mild conditions.
Comments: arXiv admin note: substantial text overlap with arXiv:2110.15318; text overlap with arXiv:2204.10607
Subjects: Machine Learning (cs.LG); Optimization and Control (math.OC)
Cite as: arXiv:2205.01438 [cs.LG]
  (or arXiv:2205.01438v4 [cs.LG] for this version)

Submission history

From: Shenglong Zhou [view email]
[v1] Tue, 3 May 2022 11:56:33 GMT (727kb)
[v2] Wed, 4 May 2022 20:04:36 GMT (579kb,D)
[v3] Wed, 9 Nov 2022 23:19:11 GMT (741kb,D)
[v4] Sun, 13 Nov 2022 21:45:14 GMT (741kb,D)

Link back to: arXiv, form interface, contact.