search

yinboc / DGP

Rethinking Knowledge Graph Propagation for Zero-Shot Learning, in CVPR 2019
MIT License
319 stars 57 forks source link

Details of SGCN #10

Closed mrzhang11 closed 5 years ago

mrzhang11 commented 5 years ago

Thanks for your contributions on this work, but I have some questions about SGCN.

In the paper, it mentions that "As DGP, our SGCN model makes use of the proposed two-stage finetuning approach", but did it use the descendant and ancestor patterns and weighting schema, like DGP as well?

From the code, it seems that SGCN is very similar to GCNZ. Their differences mainly are the number of hidden layers, training epochs and implementation details, is it corret?

Thanks very much!

yinboc commented 5 years ago

Thanks for your question.

Yes, SGCN is a single-hidden-layer GCN with finetuning, you can check the appendix in our paper for ablation studies.

mrzhang11 commented 5 years ago

Okay, thanks for your quick reply!