Skip to content

GitLab

Closed
Created by Miquel Duran-Frigola@mduranDeveloper

Augment sampling in node2vec

This is just an issue to consider, in case node2vec predict() method does not perform well.

We can run node2vec multiple times (with different random seeds) so that we have more "samples".

The only thing to take care of is that in the train/test validation split one needs to ensure that duplicates are not spread across train and test sets.