Open dongZheX opened 2 years ago
Hi! The goal of GIN is more of trying to capture the topological structure of nodes' neighborhood rather than smoothing the node features. The summation aggregation of GIN is suitable for capturing the topology, but if you apply it to social networks (some nodes have very high node degrees), you may suffer from exploding node feature problem due to the sum.
Hello, excellent job!
I have a little question about GIN. Does GIN make nodes smoother and smoother like GCN? If not, does node feature represent the structure around this node?
Thanks.