Improving Molecular Graph Neural Network Explainability with Orthonormalization and Induced Sparsity

Open in new window