Explaining GNN Explanations with Edge Gradients
Authors:
- Jesse HeJHJesse He
- Akbar RafieyARAkbar Rafiey
- Gal MishneGMGal Mishne
- Yusu WangYWYusu Wang
Preprint
Abstract
In recent years, the remarkable success of graph neural networks (GNNs) on graph-structured data has prompted a surge of methods for explaining GNN predictions. However, the state-of-the-art for GNN explainability remains in flux. Different comparisons find mixed results for different methods, with many explainers struggling on more complex GNN architectures and tasks. This presents an urgent need for a more careful theoretical analysis of competing GNN explanation methods. In this work we take a closer look at GNN explanations in two different settings: input-level explanations, which produce explanatory subgraphs of the input graph, and layerwise explanations, which produce explanatory subgraphs of the computation graph. We establish the first theoretical connections between the popular perturbation-based and classical gradient-based methods, as well as point out connections between other recently proposed methods. At the input level, we demonstrate conditions under which GNNExplainer can be approximated by a simple heuristic based on the sign of the edge gradients. In the layerwise setting, we point out that edge gradients are equivalent to occlusion search for linear GNNs. Finally, we demonstrate how our theoretical results manifest in practice with experiments on both synthetic and real datasets.Keywords
All Related Versions
- Version 1, 2025-08-01, ArXiv
- Published version: , 884.
This publication has 8 references indexed in Scilit:
- Explaining the Explainers in Graph Neural Networks: a Comparative StudyACM Computing Surveys, 2025
- Evaluating explainability for graph neural networksScientific Data, 2023
- GraphLIME: Local Interpretable Model Explanations for Graph Neural NetworksIEEE Transactions on Knowledge and Data Engineering, 2022
- When Comparing to Ground Truth is Wrong2021
- Explainability Methods for Graph Convolutional Neural Networks2019
- Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization2017
- Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification2015
- On Pixel-Wise Explanations for Non-Linear Classifier Decisions by Layer-Wise Relevance PropagationPLOS ONE, 2015