Questions about paper

[ligeng0197]

Hi, Huang! Glad to read your inspiring paper, and I have some questions about the paper details.

1. The paper summarizes the graph meta learning problem into 3 categories, and declaims meta-gnn is not appliable for multiple graph cases. However, meta-gnn just uses maml on gcn or sgc, which doesn't limit its application ocassions. So I think you can extend the algorithm to multiple graph cases just by sampling tasks on different graphs and train together.
2. Do you have comparing experiments on citeseer or cora datasets which are used more commonly? I see the meta-gnn result in table (3) is quite low which seems very strange comparing to original paper outstanding results. Thanks!