Skip to main content
Cornell University
We gratefully acknowledge support from the Simons Foundation, member institutions, and all contributors. Donate
arxiv logo > cs > arXiv:2111.08202

Help | Advanced Search

arXiv logo
Cornell University Logo

quick links

  • Login
  • Help Pages
  • About

Computer Science > Machine Learning

arXiv:2111.08202 (cs)
[Submitted on 16 Nov 2021 (v1), last revised 13 Mar 2022 (this version, v4)]

Title:Learn Locally, Correct Globally: A Distributed Algorithm for Training Graph Neural Networks

Authors:Morteza Ramezani, Weilin Cong, Mehrdad Mahdavi, Mahmut T. Kandemir, Anand Sivasubramaniam
View a PDF of the paper titled Learn Locally, Correct Globally: A Distributed Algorithm for Training Graph Neural Networks, by Morteza Ramezani and 4 other authors
View PDF
Abstract:Despite the recent success of Graph Neural Networks (GNNs), training GNNs on large graphs remains challenging. The limited resource capacities of the existing servers, the dependency between nodes in a graph, and the privacy concern due to the centralized storage and model learning have spurred the need to design an effective distributed algorithm for GNN training. However, existing distributed GNN training methods impose either excessive communication costs or large memory overheads that hinders their scalability. To overcome these issues, we propose a communication-efficient distributed GNN training technique named $\text{Learn Locally, Correct Globally}$ (LLCG). To reduce the communication and memory overhead, each local machine in LLCG first trains a GNN on its local data by ignoring the dependency between nodes among different machines, then sends the locally trained model to the server for periodic model averaging. However, ignoring node dependency could result in significant performance degradation. To solve the performance degradation, we propose to apply $\text{Global Server Corrections}$ on the server to refine the locally learned models. We rigorously analyze the convergence of distributed methods with periodic model averaging for training GNNs and show that naively applying periodic model averaging but ignoring the dependency between nodes will suffer from an irreducible residual error. However, this residual error can be eliminated by utilizing the proposed global corrections to entail fast convergence rate. Extensive experiments on real-world datasets show that LLCG can significantly improve the efficiency without hurting the performance.
Comments: The Tenth International Conference on Learning Representations (ICLR 2022)
Subjects: Machine Learning (cs.LG)
Cite as: arXiv:2111.08202 [cs.LG]
  (or arXiv:2111.08202v4 [cs.LG] for this version)
  https://doi.org/10.48550/arXiv.2111.08202
arXiv-issued DOI via DataCite

Submission history

From: Morteza Ramezani [view email]
[v1] Tue, 16 Nov 2021 03:07:01 UTC (1,293 KB)
[v2] Wed, 17 Nov 2021 17:06:22 UTC (1,251 KB)
[v3] Tue, 7 Dec 2021 16:55:20 UTC (1,252 KB)
[v4] Sun, 13 Mar 2022 14:51:02 UTC (1,246 KB)
Full-text links:

Access Paper:

    View a PDF of the paper titled Learn Locally, Correct Globally: A Distributed Algorithm for Training Graph Neural Networks, by Morteza Ramezani and 4 other authors
  • View PDF
  • TeX Source
  • Other Formats
view license
Current browse context:
cs
< prev   |   next >
new | recent | 2021-11
Change to browse by:
cs.LG

References & Citations

  • NASA ADS
  • Google Scholar
  • Semantic Scholar

DBLP - CS Bibliography

listing | bibtex
Weilin Cong
Mehrdad Mahdavi
Mahmut T. Kandemir
Anand Sivasubramaniam
a export BibTeX citation Loading...

BibTeX formatted citation

×
Data provided by:

Bookmark

BibSonomy logo Reddit logo

Bibliographic and Citation Tools

Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)

Code, Data and Media Associated with this Article

alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)

Demos

Replicate (What is Replicate?)
Hugging Face Spaces (What is Spaces?)
TXYZ.AI (What is TXYZ.AI?)

Recommenders and Search Tools

Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
IArxiv Recommender (What is IArxiv?)
  • Author
  • Venue
  • Institution
  • Topic

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)
  • About
  • Help
  • contact arXivClick here to contact arXiv Contact
  • subscribe to arXiv mailingsClick here to subscribe Subscribe
  • Copyright
  • Privacy Policy
  • Web Accessibility Assistance
  • arXiv Operational Status
    Get status notifications via email or slack