Can we apply edge weights in GAT ? I’m using GCN on a citation coupling graph with edge weight indicating coupling strength. I want to use GAT for better aggregation of node attribute(generated by BERT using paper abstract), while both coupling strength and semantic imformation are important in this graph, how to implement them together?
Maybe EGATConv — DGL 0.8.0post1 documentation? (see also GitHub - labstructbioinf/EdgeGat: graph neural network layer for Deep Graph Library with pytorch backend).
Best,
Staszek
1 Like
This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.