FIGNN: Feature-Specific Interpretability for Graph Neural Network Surrogate Models

Open in new window