chakib401 / sagsc Goto Github PK

We wish to learn the embeddings of nodes in weighted biological networks and then perform clustering. We found that most graph representation learning methods are not designed for weighted networks. We were lucky to find that SAGSC can be used for clustering weighted networks. However, we found that in the node embeddings learned by SAGSC, although nodes in the same cluster are closely connected, there are still some cluttered points. We hope to further optimize and look forward to your suggestions and help.

Visualization of original node features:

Visualization of node embeddings from SASGC：