Zheyu Zhang

Thank you. But why would you divide the KL term by N^2 when it is the sum of N terms? Would it make more sense to divide it by N...

Thanks for the suggestions:)

Hi @tkipf, Have you tried to output the reconstructed adjacency matrix based on the training data? I have found the reconstruction deviated a lot from the original matrix (for the...

Hi @tkipf, Thank you for your suggestion, I have tried on various thresholds. But it seems the reconstructed adjacency matrix is still very noisy even when the threshold is above...

Thanks for clarifying and your further suggestions.