ICML 2019oral456 citations

On the Convergence and Robustness of Adversarial Training

Yisen Wang, Xingjun Ma, James Bailey, Jinfeng Yi, Bowen Zhou, Quanquan Gu

Abstract

Improving the robustness of deep neural networks (DNNs) to adversarial examples is an important yet challenging problem for secure deep learning. Across existing defense techniques, adversarial training with Projected Gradient Decent (PGD) is amongst the most effective. Adversarial training solves a min-max optimization problem, with the

BibTeX
@InProceedings{pmlr-v97-wang19i,
  title = 	 {On the Convergence and Robustness of Adversarial Training},
  author =       {Wang, Yisen and Ma, Xingjun and Bailey, James and Yi, Jinfeng and Zhou, Bowen and Gu, Quanquan},
  booktitle = 	 {Proceedings of the 36th International Conference on Machine Learning},
  pages = 	 {6586--6595},
  year = 	 {2019},
  editor = 	 {Chaudhuri, Kamalika and Salakhutdinov, Ruslan},
  volume = 	 {97},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {09--15 Jun},
  publisher =    {PMLR},
  pdf = 	 {http://proceedings.mlr.press/v97/wang19i/wang19i.pdf},
  url = 	 {https://proceedings.mlr.press/v97/wang19i.html},
  abstract = 	 {Improving the robustness of deep neural networks (DNNs) to adversarial examples is an important yet challenging problem for secure deep learning. Across existing defense techniques, adversarial training with Projected Gradient Decent (PGD) is amongst the most effective. Adversarial training solves a min-max optimization problem, with the
On the Convergence and Robustness of Adversarial Training · ICML 2019