Learning-Invariant-Representations-and-Risks icon indicating copy to clipboard operation
Learning-Invariant-Representations-and-Risks copied to clipboard

Published 20 hours ago • verified publisher icon Luodian

Question about the 1% setting in the Office-Home dataset

Open abhayraw1 opened this issue 3 years ago • 0 comments

I had a doubt about using 1% data in the Office-Home dataset. As the dataset is very small, (~ 4000 samples per domain with 65 classes) , using 1% of the data means that some of the samples might not be in the training set. How is this taken care of? I know I could go through the code myself but I think it's something I'd rather ask.

Thanks.!

abhayraw1 avatar Dec 13 '21 16:12 abhayraw1