Giters

balcilar / Spectral-Designed-Graph-Convolutions

Codes for "Bridging the Gap Between Spectral and Spatial Domains in Graph Neural Networks" paper

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Question on the num_feat for ENZYMES dataset

lizaitang opened this issue · comments

commented

Dear Author, your code really helps a lot. I have a question that seems that node_feature in original dataset is 18, while in the mat you provided is 21, how do you transfer it?

commented

Thanks for your attention to our paper and code bases. The extra 3 features are the one-hot coding of node type. As it can be seen on the table 4 in the paper (https://arxiv.org/pdf/2003.11702.pdf) there are 18 continuous features and 3 type of nodes. So we encode node type as a categorical features of the nodes. Hope this explanation works for you.

commented

Thank you so much，so where should we add the one hot vector? right after the 18 features？