Node Classification Task

YH-UtMSB / sigvae-torch

A Pytorch implementation of Semi-implicit-graph-variational-auto-encoders

14 stars 5 forks source link

Node Classification Task #2

Open rd27995 opened 3 years ago

rd27995 commented 3 years ago

Hello,

Can you provide some explanation on how to use this code repository for node classification task?

YH-UtMSB commented 3 years ago

Hi, thanks for the feedback!

This repo only re-produced the graph reconstruction function from the original paper, as for node classification, according to section 5.4, I think the adjustments they made to enable node classification is to append an output layer to the latent node representation (denoted by Z in the paper) or its mean (denoted by \mu), and use a softmax loss upon the labeled nodes.