GOAt: Explaining Graph Neural Networks via Graph Output Attribution

TL;DR

GOAt introduces an analytical approach to explain GNN decisions by attributing output importance to input features, outperforming existing methods in fidelity, discriminability, and stability.

Contribution

The paper presents GOAt, a novel analytical method for explaining GNNs by attributing output to input features without auxiliary models, enhancing interpretability.

Findings

Outperforms state-of-the-art GNN explainers in fidelity

Exhibits stronger discriminability in explanations

Demonstrates greater stability across similar samples

Abstract

Understanding the decision-making process of Graph Neural Networks (GNNs) is crucial to their interpretability. Most existing methods for explaining GNNs typically rely on training auxiliary models, resulting in the explanations remain black-boxed. This paper introduces Graph Output Attribution (GOAt), a novel method to attribute graph outputs to input graph features, creating GNN explanations that are faithful, discriminative, as well as stable across similar samples. By expanding the GNN as a sum of scalar products involving node features, edge features and activation patterns, we propose an efficient analytical method to compute contribution of each node or edge feature to each scalar product and aggregate the contributions from all scalar products in the expansion form to derive the importance of each node and edge. Through extensive experiments on synthetic and real-world data, we show that our method not only outperforms various state-ofthe-art GNN explainers in terms of the commonly used fidelity metric, but also exhibits stronger discriminability, and stability by a remarkable margin.