Replies: 1 comment
-
Inner product is certainly the most common since it allows you to do Maximum Inner Product Search (MIPS) on node embeddings. You can also use other k-NN tools, e.g., based on Euclidean distance). Alternatively, you can have a parametrized decoder, e.g., an MLP on source and destination node embeddings that maps to a link probability, but this is expensive to compute for O(N^2) pairs. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Dear pyg users and developers,
Is there a way other than inner product to reconstruct the node-to-edge representation here?
If so, please let me know the paper or the exact method.
Beta Was this translation helpful? Give feedback.
All reactions