I see in your paper, you use sum of more than one adjacency matrix(like different power of the origin matrix), but there is no such operation in code implementation, is there any reason, and does it affect performance?
I really like your paper and was more interested in it, so I took into multiplex datasets, and for these two datasets in table 4, it is written that both of them have 2 classes. On the other hand, you have cited DMGI paper as a source of your datasets, but DMGI paper has 3 classes for each of them. Maybe I got something wrong and clarification would help. So, could you please help me with this? Why do you have two classes instead of three and how did you implement this?