Evaluating Neighbor Explainability for Graph Neural Networks

Oscar Llorente, Rana Fawzy, Jared Keown, Michal Horemuz, Péter Vaderna +4 more
11/14/2023
cs.LGcs.AI

No official code found. Best implementation: EricssonResearch/gnn-neighbors-xai (4 stars, last commit 2023-11-15).

Abstract

Explainability in Graph Neural Networks (GNNs) is a new field growing in the last few years. In this publication we address the problem of determining how important is each neighbor for the GNN when classifying a node and how to measure the performance for this specific task. To do this, various known explainability methods are reformulated to get the neighbor importance and four new metrics are presented. Our results show that there is almost no difference between the explanations provided by gradient-based techniques in the GNN domain. In addition, many explainability techniques failed to identify important neighbors when GNNs without self-loops are used.

View on arXivView PDF

Code Implementations(1)

EricssonResearch/gnn-neighbors-xaiCommunity64%

Official implementation of “Evaluating Neighbor Explainability for Graph Neural Networks”

40Python, Jupyter NotebookAug 16, 20232 years ago

Similar Papers

A Survey on Explainability of Graph Neural Networks

88% match2023

Do graph neural network states contain graph properties?

85% match2024

Parallelizing Node-Level Explainability in Graph Neural Networks

85% match2026

Explaining Graph Neural Networks for Node Similarity on Graphs

84% match2024

GNNAnatomy: Rethinking Model-Level Explanations for Graph Neural Networks

84% match2024

Self-Explainable Graph Neural Networks for Link Prediction

83% match2023

Cite this paper

@article{llorente2023evaluating,
  title  = {Evaluating Neighbor Explainability for Graph Neural Networks},
  author = {Oscar Llorente and Rana Fawzy and Jared Keown and Michal Horemuz and Péter Vaderna and Sándor Laki and Roland Kotroczó and Rita Csoma and János Márk Szalai-Gindl},
  year   = {2023},
  eprint = {2311.08118},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2311.08118v3}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2311.08118.svg)](https://codeofpaper.com/paper/2311.08118)

Discussion