Layer-wise training for self-supervised learning on graphs

Oscar Pina, Verónica Vilaplana

End-to-end training of graph neural networks (GNN) on large graphs presents several memory and computational challenges, and limits the application to shallow architectures as dept…