Graph Convolutional Neural Networks Sensitivity under Probabilistic Error Model

Xinjue Wang, Esa Ollila*, Sergiy A. Vorobyov

*Corresponding author for this work

Research output: Contribution to journalArticleScientificpeer-review

9 Downloads (Pure)

Abstract

Graph Neural Networks (GNNs), particularly Graph Convolutional Neural Networks (GCNNs), have emerged as pivotal instruments in machine learning and signal processing for processing graph-structured data. This paper proposes an analysis framework to investigate the sensitivity of GCNNs to probabilistic graph perturbations, directly impacting the graph shift operator (GSO). Our study establishes tight expected GSO error bounds, which are explicitly linked to the error model parameters, and reveals a linear relationship between GSO perturbations and the resulting output differences at each layer of GCNNs. This linearity demonstrates that a single-layer GCNN maintains stability under graph edge perturbations, provided that the GSO errors remain bounded, regardless of the perturbation scale. For multilayer GCNNs, the dependency of system's output difference on GSO perturbations is shown to be a recursion of linearity. Finally, we exemplify the framework with the Graph Isomorphism Network (GIN) and Simple Graph Convolution Network (SGCN). Experiments validate our theoretical derivations and the effectiveness of our approach.

Original languageEnglish
Pages (from-to)788-803
Number of pages16
JournalIEEE Transactions on Signal and Information Processing over Networks
Volume10
Early online date2024
DOIs
Publication statusPublished - 2024
MoE publication typeA1 Journal article-refereed

Keywords

  • Sensitivity analysis
  • graph convolutional neural network
  • graph shift operator
  • structural perturbation
  • sensitivity analysis
  • Graph convolutional neural network

Fingerprint

Dive into the research topics of 'Graph Convolutional Neural Networks Sensitivity under Probabilistic Error Model'. Together they form a unique fingerprint.

Cite this