search

matenure / FastGCN

The sample codes for our ICLR18 paper "FastGCN: Fast Learning with Graph Convolutional Networks via Importance Sampling""
519 stars 111 forks source link

What are the pubmed-original_inductive_FastGCN.py and pubmed-original_transductive_FastGCN.py meaning? #16

Open kugwzk opened 5 years ago

kugwzk commented 5 years ago

Hi~ @matenure I want to change the some code to pytorch version, so I have some question about the use of inductive file and the transductive file. And if I have an inductive task for GCN, but the origin GCN can't use the mini-batch, how could I add the code so that it can be used with mini-batch? Thanks a lot.

matenure commented 5 years ago

Thanks for transferring it to pytorch.

As to the minibatch, I suggest developing only one minibatch version for GCN, so that if you do not want to use minibatch, i.e. the original version, you just need to define the batch size as the total number of nodes in the graph.