Repository logo

Infoscience

  • English
  • French
Log In
Logo EPFL, École polytechnique fédérale de Lausanne

Infoscience

  • English
  • French
Log In
  1. Home
  2. Academic and Research Output
  3. Conferences, Workshops, Symposiums, and Seminars
  4. Spectrally approximating large graphs with smaller graphs
 
conference paper

Spectrally approximating large graphs with smaller graphs

Loukas, Andreas  
•
Vandergheynst, Pierre  
July 10, 2018
Proceedings of International Conference in Machine Learning 2018
International Conference in Machine Learning (ICML)

How does coarsening affect the spectrum of a general graph? We provide conditions such that the principal eigenvalues and eigenspaces of a coarsened and original graph Laplacian matrices are close. The achieved approximation is shown to depend on standard graph-theoretic properties, such as the degree and eigenvalue distributions, as well as on the ratio between the coarsened and actual graph sizes. Our results carry implications for learning methods that utilize coarsening. For the particular case of spectral clustering, they imply that coarse eigenvectors can be used to derive good quality assignments even without refinement---this phenomenon was previously observed, but lacked formal justification.

  • Files
  • Details
  • Metrics
Loading...
Thumbnail Image
Name

1802.07510.pdf

Access type

openaccess

Size

1.46 MB

Format

Adobe PDF

Checksum (MD5)

d189033d4ef5c143213e8b6bdeed3903

Logo EPFL, École polytechnique fédérale de Lausanne
  • Contact
  • infoscience@epfl.ch

  • Follow us on Facebook
  • Follow us on Instagram
  • Follow us on LinkedIn
  • Follow us on X
  • Follow us on Youtube
AccessibilityLegal noticePrivacy policyCookie settingsEnd User AgreementGet helpFeedback

Infoscience is a service managed and provided by the Library and IT Services of EPFL. © EPFL, tous droits réservés