smoothing

smoothing copied to clipboard

No clamp for input after Gaussian data augumentation?

1

``` # augment inputs with noise inputs = inputs + torch.randn_like(inputs, device='cuda') * noise_sd ```` Line 109 in `train.py`. If there is no `torch.clamp()` after line 109, it is possible...

xinghua-qu

Is it possible for the smoothed classifier to completely abstain on test set?

2

@jmcohen Hi, thanks for releasing the code. If you don't mind me asking, I'm trying to understand if its possible for a smooth classifier trained using randomised smoothing to completely...

kirk86

attack

1

Would like to ask how I will attack your method using my own generated counter samples? Looking forward to your reply, appreciate it!

tanlingp

Hi the paper doesn't explicitly mention what the white line represents. Is that just the range for which R can be in?  Since we are using estimated values for...

jyaacoub

How to draw the dashed black line in fig6

hh-wq