xixi

关于RSD论文中 特征矩阵的F范数

4

您好 请问RSD论文中figure1（b）具体是如何得到的？文中 “we use L2 regularization to change the Frobenius norm of feature matrix. "对于这句话不太理解，可以简单解释一下吗？非常感谢

about domain acc

1

Thanks for your sharing. I try to apply this method (cdan_sdat.py) to UDA for an image regression task. But the domain acc keeps 100% with no decline. Can you give...

Hi, thanks for your sharing. In "anchor.py", it seems that no Gauss map is used in non-local operation and the calculation process is inconsistent with the paper. Could you please...

Thanks for you excellent work. I wonder when will you add those scripts for PACS and imagenet ？

Missing file

1

Hello！Thanks for sharing. "imi_dataset_complex.py" and "transformer_dataset.py" are not included in this repo. Could you please upload them？Thanks a lot!