GNNExplainer: Generating Explanations for Graph Neural Networks

• https://arxiv.org/pdf/1903.03894.pdf
• Rex Ying, …, Marinka Zitnik, Jure Leskovec

Graph neural network, Explainability

The key idea is to find a subgraph $G_S$ and their features $X_S$ that are important for the GNN’s prediction $\hat{y}$. This can be formulated as the Mutual information