Node embedding for discoonected nodes

Questions
#1

I have 68 small graphs, each has some nodes that are disconnected (it doesn’t have any incoming/outgoing edge). The end goal is to classify each node (Node Classification) based on its class (there is only one class for each node, and features for edges and nodes are not available), therefore, the disconnected nodes must also be considered in the graph dataset.

My question is:

Can Node2vec or other Embedding algorithms like Struc2vec operate on completely disconnected nodes from the graph?

#2

For graph embedding methods like Node2Vec, they cannot yield meaningful embeddings for isolated nodes. For GNN methods, they cannot yield meaningful embeddings for isolated nodes without input node features.

#3

This topic was automatically closed 30 days after the last reply. New replies are no longer allowed.