Building powerful and equivariant graph neural networks with structural message-passing

Open in new window