Why Does Dropping Edges Usually Outperform Adding Edges in Graph Contrastive Learning?

Open in new window