Learning with a Strong Adversary
Ruitong Huang
and
Bing Xu
and
Dale Schuurmans
and
Csaba Szepesvari
arXiv e-Print archive - 2015 via Local arXiv
Keywords:
cs.LG
First published: 2015/11/10 (9 years ago) Abstract: The robustness of neural networks to intended perturbations has recently
attracted significant attention. In this paper, we propose a new method,
\emph{learning with a strong adversary}, that learns robust classifiers from
supervised data. The proposed method takes finding adversarial examples as an
intermediate step. A new and simple way of finding adversarial examples is
presented and experimentally shown to be efficient. Experimental results
demonstrate that resulting learning method greatly improves the robustness of
the classification models produced.