Evaluating Neighbor Explainability for Graph Neural Networks
This addresses explainability for GNN users, but it is incremental as it reformulates existing methods and introduces new metrics without major breakthroughs.
The paper tackled the problem of evaluating neighbor importance in Graph Neural Networks (GNNs) for node classification, finding that gradient-based techniques provide similar explanations and many methods fail with GNNs without self-loops.
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.