@misc{indiciae5de2915085f5,
  title = {Lagrangian Objective Function Leads to Improved Unforeseen Attack Generalization in Adversarial Training},
  author = {Mohammad Azizmalayeri and Mohammad Hossein Rohban},
  year = {2021},
  url = {https://arxiv.org/abs/2103.15385},
  note = {Source identifier: 2103.15385}
}