Great job.I am trying to incorporate it into my work.
I noticed that you used JT-VAE. It may generate such subgraphs
cliques = [[0, 1], [9, 10], [9, 16], [1, 2, 3, 4, 5, 6, 7, 8, 9], [11, 12, 13, 14, 15, 10], [17, 18, 19, 20, 21, 16], [9]].
I noticed your code ‘atom_sum_embedding = scatter(atom_embedding, batch.map, dim=0, reduce="mean")’
When using PS-VAE, it works very well because each node corresponds to a subgraph. My problem is that a node in JT-VAE may correspond to multiple subgraphs. I don't know how to aggregate them.
I hope my description of the problem is clear enough. Since I am new to this field, I may have overlooked some details. I look forward to your response. Best wishes.
Great job.I am trying to incorporate it into my work.
I noticed that you used JT-VAE. It may generate such subgraphs
cliques = [[0, 1], [9, 10], [9, 16], [1, 2, 3, 4, 5, 6, 7, 8, 9], [11, 12, 13, 14, 15, 10], [17, 18, 19, 20, 21, 16], [9]].
I noticed your code ‘atom_sum_embedding = scatter(atom_embedding, batch.map, dim=0, reduce="mean")’
When using PS-VAE, it works very well because each node corresponds to a subgraph. My problem is that a node in JT-VAE may correspond to multiple subgraphs. I don't know how to aggregate them.
I hope my description of the problem is clear enough. Since I am new to this field, I may have overlooked some details. I look forward to your response. Best wishes.