Improving Graph Neural Network Generalization by Learning Personalized Scoping under Heterophily

Accurately predicting the optimal depth of a given Graph Neural Network (GNN) architecture for each node can significantly enhance the generalization capability of GNNs on heterophilous graphs.