Provably Robust Deep Learning via Adversarially Trained Smoothed Classifiers — mentioned by orange3xchicken

Fri, 06 Aug 2021 15:28:42 GMT

It turns out that a similar technique, where you basically apply noise multiple times to a single image, and average predictions over all noisy images- equivalent to convolving your nn with Gaussian noise yields near state of the art bounds on provable robustness (under a specific class of attacks). The issue is the magnitude of noise you need in order to get practically robust networks is quite large relative to the data you are dealing with. https://arxiv.org/abs/1902.02918 https://arxiv.org/abs/1906.04584

Paper and discussion history. A mention is not an endorsement.

Provably Robust Deep Learning via Adversarially Trained Smoothed Classifiers — HN submission

Thu, 13 Jun 2019 18:14:27 GMT

Provably Robust Deep Learning

2 points; 0 discussion replies on HN. Paper and source history. A submission is not an endorsement.

Provably Robust Deep Learning via Adversarially Trained Smoothed Classifiers · HackerPapers discussions

Provably Robust Deep Learning via Adversarially Trained Smoothed Classifiers — mentioned by orange3xchicken

Provably Robust Deep Learning via Adversarially Trained Smoothed Classifiers — HN submission