You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
First,thank you very much,You have done a very good job.
I don't unstand the following content in your paper
Figure 4: Magnitude of feature activations, sorted by descending value, and averaged over all test samples. A standard ResNet18 is compared with a ResNet18 trained with cutout at three different depths.
could your give me some tips how do you generate these figures?
and I don't know the means of the abscissa and ordinate values.
I need you help.thank you.
The text was updated successfully, but these errors were encountered:
Hello, I've attached an old and messy script that I think was used to create the plots from the paper. You'll have to change the file extension from .txt back to .py if you want to use it.
In summary, we take the maximum value from each feature map, sort the values in the channel dimension, calculate the average across the batch dimension, and then sort again across channels.
First,thank you very much,You have done a very good job.
I don't unstand the following content in your paper
Figure 4: Magnitude of feature activations, sorted by descending value, and averaged over all test samples. A standard ResNet18 is compared with a ResNet18 trained with cutout at three different depths.
could your give me some tips how do you generate these figures?
and I don't know the means of the abscissa and ordinate values.
I need you help.thank you.
The text was updated successfully, but these errors were encountered: