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

Download:

Current browse context:

cs.SI

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 > Social and Information Networks

Title: Low-Rank Projections of GCNs Laplacian

Authors: Nathan Grinsztajn (Scool), Philippe Preux (Scool), Edouard Oyallon (MLIA)
Abstract: In this work, we study the behavior of standard models for community detection under spectral manipulations. Through various ablation experiments, we evaluate the impact of bandpass filtering on the performance of a GCN: we empirically show that most of the necessary and used information for nodes classification is contained in the low-frequency domain, and thus contrary to images, high frequencies are less crucial to community detection. In particular, it is sometimes possible to obtain accuracies at a state-of-the-art level with simple classifiers that rely only on a few low frequencies.
Subjects: Social and Information Networks (cs.SI); Machine Learning (cs.LG)
Journal reference: ICLR 2021 Workshop GTRL, 2021, Online, France
Cite as: arXiv:2106.07360 [cs.SI]
  (or arXiv:2106.07360v1 [cs.SI] for this version)

Submission history

From: Edouard Oyallon [view email]
[v1] Fri, 4 Jun 2021 09:54:26 GMT (1109kb,D)

Link back to: arXiv, form interface, contact.