Generative Explanations for Graph Neural Network: Methods and Evaluations

TL;DR

This paper reviews generative explanation methods for Graph Neural Networks, proposing a unified optimization framework, and evaluates their performance, efficiency, and generalizability.

Contribution

It introduces a unified optimization objective for generative GNN explanations and analyzes various methods under this framework.

Findings

Unified explanation objective reveals shared characteristics of methods.

Empirical evaluations compare explanation performance and efficiency.

Highlights limitations and potential improvements in GNN explainability.

Abstract

Graph Neural Networks (GNNs) achieve state-of-the-art performance in various graph-related tasks. However, the black-box nature often limits their interpretability and trustworthiness. Numerous explainability methods have been proposed to uncover the decision-making logic of GNNs, by generating underlying explanatory substructures. In this paper, we conduct a comprehensive review of the existing explanation methods for GNNs from the perspective of graph generation. Specifically, we propose a unified optimization objective for generative explanation methods, comprising two sub-objectives: Attribution and Information constraints. We further demonstrate their specific manifestations in various generative model architectures and different explanation scenarios. With the unified objective of the explanation problem, we reveal the shared characteristics and distinctions among current methods, laying the foundation for future methodological advancements. Empirical results demonstrate the advantages and limitations of different explainability approaches in terms of explanation performance, efficiency, and generalizability.