Hadisalman / smoothing-adversarial

Code for our NeurIPS 2019 *spotlight* "Provably Robust Deep Learning via Adversarially Trained Smoothed Classifiers"

Home Page:https://arxiv.org/abs/1906.04584

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

InputCenterLayer?

shudong-zhang opened this issue · comments

The difference between InputCenterLayer and NormalizeLayer is simply that InputCenterLayer is not divided by the sds. What are the benefits of doing this?

Apologies for the late reply, for some reason I didn't notice the issue previously. We didn't find any benefits from doing either. We experimented with both, and eventually we sticked with one. They both give very similar results.