Graph Neural Networks (GNNs) have emerged as a versatile framework for learning from graph-structured data by propagating and aggregating node and edge features along the topology. Methodological ...