ProtGNN: Towards Self-Explaining Graph Neural Networks

Despite the recent progress in Graph Neural Networks (GNNs), it remains challenging to explain predictions made by GNNs. Existing explanation methods mainly focus on post-hoc explanations where another explanatory model is employed provide for a trained GNN. The fact that fail reveal original reasoning process of GNNs raises need building with built-in interpretability. In this work, we propose Prototype Network (ProtGNN), which combines prototype learning and provides new perspective ProtGNN, are naturally derived from case-based actually used during classification. prediction ProtGNN obtained comparing inputs few learned prototypes latent space. Furthermore, better interpretability higher efficiency, novel conditional subgraph sampling module incorporated indicate part input graph most similar each ProtGNN+. Finally, evaluate our method wide range datasets perform concrete case studies. Extensive results show ProtGNN+ can inherent while achieving accuracy par non-interpretable counterparts.