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: In-Network Learning: Distributed Training and Inference in Networks

Abstract: It is widely perceived that leveraging the success of modern machine learning techniques to mobile devices and wireless networks has the potential of enabling important new services. This, however, poses significant challenges, essentially due to that both data and processing power are highly distributed in a wireless network. In this paper, we develop a learning algorithm and an architecture that make use of multiple data streams and processing units, not only during the training phase but also during the inference phase. In particular, the analysis reveals how inference propagates and fuses across a network. We study the design criterion of our proposed method and its bandwidth requirements. Also, we discuss implementation aspects using neural networks in typical wireless radio access; and provide experiments that illustrate benefits over state-of-the-art techniques.
Comments: Submitted to the IEEE Journal on Selected Areas in Information Theory (JSAIT) Special Issue on Information Theoretic Foundations of Future Communication Systems. arXiv admin note: substantial text overlap with arXiv:2104.14929
Subjects: Machine Learning (cs.LG); Information Theory (cs.IT); Machine Learning (stat.ML)
Cite as: arXiv:2107.03433 [cs.LG]
  (or arXiv:2107.03433v2 [cs.LG] for this version)

Submission history

From: Matei Moldoveanu [view email]
[v1] Wed, 7 Jul 2021 18:35:08 GMT (7920kb,D)
[v2] Fri, 17 Sep 2021 10:48:38 GMT (7920kb,D)

Link back to: arXiv, form interface, contact.