Dongsheng Luo

A new method improves understanding of predictions in graph regression tasks.

New method improves GNN explainability using proxy graphs.

This study improves GNN performance through effective graph explanations.

New findings challenge the idea that classification and explanation robustness are linked.