Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Latest commit 80f3207 Dec 18, 2018
Type Name Latest commit message Commit time
Failed to load latest commit information. Update Dec 18, 2018


PROVEN is general and efficient framework to PRObabilistic VErify Neural network robustness -- it certifies the probability that the classifier's top-1 prediction can't be altered under any constrained Lp norm perturbation to a given input.

  • PROVEN is general and efficient: it can be directly built on top of existing neural network robustness verification algorithms including Fast-Lin, CROWN and the latest CNN-Cert to efficiently certify robustness of convolutional neural networks (CNNs) with little computation overhead.


We are preparing for the code release, please stay tuned!

You can’t perform that action at this time.