The adj after dropedge is asymmetrical？ #11

Activity

1. wdzhong commented on Sep 28, 2020

   

   wdzhong

   on Sep 28, 2020

   More actions

   I have the same doubt. The paper does not mention anything about directed graph or undirected graph. The paper only mentions that the edges to be dropped are randomly selected.
2. guixianjin commented on Nov 27, 2020

   

   guixianjin

   on Nov 27, 2020

   Last edited by guixianjin

   More actions

   I also have the same doubt. I find the code cannot guarantee the symmetry of the sampled adj matrix. Thus the symmetric Normalization $\tilde{D}^{-1/2} \tilde{A} \tilde{D}^{-1/2}$ may be a problem when $\tilde{A}$ may not be symmetric?