Thanks for your great work, but I have some questions about the training batch size.
Must data be entered all at once?
Can't we have batch_size?
If we have batchsize, what do we do with the adjacency matrix?
Because GCN is trained with full-batch, so SDCN is also full-batch. If you want to use mini-batch, you can try the sampling-based GNNs, e.g., GraphSAGE.
Thanks for your great work, but I have some questions about the training batch size. Must data be entered all at once? Can't we have batch_size? If we have batchsize, what do we do with the adjacency matrix?