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: GIDN: A Lightweight Graph Inception Diffusion Network for High-efficient Link Prediction

Abstract: In this paper, we propose a Graph Inception Diffusion Networks(GIDN) model. This model generalizes graph diffusion in different feature spaces, and uses the inception module to avoid the large amount of computations caused by complex network structures. We evaluate GIDN model on Open Graph Benchmark(OGB) datasets, reached an 11% higher performance than AGDN on ogbl-collab dataset.
Subjects: Machine Learning (cs.LG); Social and Information Networks (cs.SI)
Cite as: arXiv:2210.01301 [cs.LG]
  (or arXiv:2210.01301v2 [cs.LG] for this version)

Submission history

From: Zixiao Wang [view email]
[v1] Tue, 4 Oct 2022 01:15:39 GMT (1269kb)
[v2] Tue, 11 Oct 2022 05:43:22 GMT (1233kb)

Link back to: arXiv, form interface, contact.