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: Sustainable Federated Learning

Abstract: Potential environmental impact of machine learning by large-scale wireless networks is a major challenge for the sustainability of future smart ecosystems. In this paper, we introduce sustainable machine learning in federated learning settings, using rechargeable devices that can collect energy from the ambient environment. We propose a practical federated learning framework that leverages intermittent energy arrivals for training, with provable convergence guarantees. Our framework can be applied to a wide range of machine learning settings in networked environments, including distributed and federated learning in wireless and edge networks. Our experiments demonstrate that the proposed framework can provide significant performance improvement over the benchmark energy-agnostic federated learning settings.
Comments: 12 pages, 1 figure
Subjects: Machine Learning (cs.LG); Information Theory (cs.IT)
Cite as: arXiv:2102.11274 [cs.LG]
  (or arXiv:2102.11274v1 [cs.LG] for this version)

Submission history

From: Basak Guler [view email]
[v1] Mon, 22 Feb 2021 18:58:47 GMT (77kb,D)

Link back to: arXiv, form interface, contact.