Skip to Main content Skip to Navigation
Conference papers

BAYESIAN NODE CLASSIFICATION FOR NOISY GRAPHS

Abstract : Graph neural networks (GNN) have been recognized as powerful tools for learning representations in graph structured data. The key idea is to propagate and aggregate information along edges of the given graph. However, little work has been done to analyze the effect of noise on their performance. By conducting a number of simulations, we show that GNN are very sensitive to the graph noise. We propose a graphassisted Bayesian node classifier which takes into account the degree of impurity of the graph, and show that it consistently outperforms GNN based classifiers on benchmark datasets, particularly when the degree of impurity is moderate to high.
Complete list of metadata

https://hal.archives-ouvertes.fr/hal-03291075
Contributor : Philippe Ciblat <>
Submitted on : Monday, July 19, 2021 - 4:15:20 PM
Last modification on : Tuesday, September 21, 2021 - 2:16:04 PM

File

ssp2021.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-03291075, version 1

Citation

Hakim Hafidi, Mounir Ghogho, Philippe Ciblat, Ananthram Swami. BAYESIAN NODE CLASSIFICATION FOR NOISY GRAPHS. IEEE Statistical Signal Processing Workshop (SSP), 2021, Rio de Janeiro (virtual), Brazil. ⟨hal-03291075⟩

Share

Metrics

Record views

49

Files downloads

36