How to use for graph level unsupervised clustering, not the node.

[PeterDeSOM]

I have millions of graph data that cannot be labeled by human efforts. I need it to be classified by unsupervised technique, not semi and not the node level. Is it possible?

[fork123aniket]

I don't think I understand your question that well. Can you please elaborate on the dataset characteristics at least from a high-level perspective, i.e. no. of graphs/nodes you have, is the graph in the dataset a property/attributed graph, and what you wanna achieve (the end prediction task (e.g. Node/Graph/Edge classification, Graph Clustering, etc.) you wanna perform using GraphNet?? Proper rephrasing of the above question might also actually help me understand your question...