search

muhanzhang / pytorch_DGCNN

PyTorch implementation of DGCNN
MIT License
369 stars 122 forks source link

About node feature #21

Closed Yfhu1103 closed 4 years ago

Yfhu1103 commented 4 years ago

Hi, thanks for the amazing work. I have a question, as far as I know, In ENZYMES, every node in every graph has features, but in your 'proteins.txt', you seemed to only use node label, I donnot know why.

muhanzhang commented 4 years ago

Hi, thanks for the question. The proteins dataset I used is a simplified one without continuous node attributes. You may download the proteins_full and use my "dortmund2txt.m" MATLAB script to generate the full dataset with continuous node attributes.

Yfhu1103 commented 4 years ago

But I mean, if node attributes are not used in graph classification, will the result be not good enough? Can this code run with considering node attributes?

muhanzhang commented 4 years ago

Probably, the results won't be as good as using node attributes. This code can certainly run with considering node attributes.

Yfhu1103 commented 4 years ago

I see, Thanks a lot！