Kernels on Graphs as Proximity Measures

Abstract : Kernels and, broadly speaking, similarity measures on graphs are extensively used in graph-based unsupervised and semi-supervised learning algorithms as well as in the link prediction problem. We analytically study proximity and distance properties of various kernels and similarity measures on graphs. This can potentially be useful for recommending the adoption of one or another similarity measure in a machine learning method. Also, we numerically compare various similarity measures in the context of spectral clustering and observe that normalized heat-type similarity measures with log modification generally perform the best.
Complete list of metadatas

Cited literature [40 references]  Display  Hide  Download

https://hal.inria.fr/hal-01647915
Contributor : Konstantin Avrachenkov <>
Submitted on : Friday, November 24, 2017 - 4:35:30 PM
Last modification on : Thursday, January 11, 2018 - 4:19:02 PM

File

Paper3WAW2017.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01647915, version 1

Collections

Citation

Konstantin Avrachenkov, Pavel Chebotarev, Dmytro Rubanov. Kernels on Graphs as Proximity Measures. Proceedings of the 14th Workshop on Algorithms and Models for the Web Graph (WAW 2017), Jun 2017, Toronto, Canada. ⟨hal-01647915⟩

Share

Metrics

Record views

338

Files downloads

571