nblt/sub-at

[CVPR 2022 oral] Subspace Adversarial Training

sub-at's Issues

Questions about PCA dimensionality

Greetings. Really solid work!
However, I've got a simple question. In train_adv_psgd.py, it seems that you do PCA on W with a (num_ckpts, len_params) shape. However, according to the implementation of sklearn.decomposition.PCA, it is the dimension of num_ckpts rather than len_params that be reduced. In my understanding, it is len_params (~0.2M) that should be reduced to realize the subspace training. Is this a bug?

Purpose of `attack_pgd` function

Thanks authors for the great work.
While reviewing the code, I am confused why overwriting the pgd attack again in utils.py though it was already imported from advertorch.attack.

Best,
Anthony

q

既然可以用在对抗训练上，那是否也可以在对抗样本生成的过程中使用子空间训练呢？