A bug in the brain_dataset.py
Closed this issue · 1 comments
Gemaxis commented
Hi,
This is excellent work, but I found a bug in brain_dataset.py.
At the function named dense_to_ind_val, torch.isnan(adj)==0 will make every existing value turn into True, causing the index fully True which means every brain network is fully connected.
yangji9181 commented
Thanks For the comments. The brain networks are indeed fully connected, which are different from most graphs in other domains.