English

Semi-Implicit Hybrid Gradient Methods with Application to Adversarial Robustness

Machine Learning 2022-02-23 v1 Artificial Intelligence Computer Vision and Pattern Recognition Optimization and Control

Abstract

Adversarial examples, crafted by adding imperceptible perturbations to natural inputs, can easily fool deep neural networks (DNNs). One of the most successful methods for training adversarially robust DNNs is solving a nonconvex-nonconcave minimax problem with an adversarial training (AT) algorithm. However, among the many AT algorithms, only Dynamic AT (DAT) and You Only Propagate Once (YOPO) guarantee convergence to a stationary point. In this work, we generalize the stochastic primal-dual hybrid gradient algorithm to develop semi-implicit hybrid gradient methods (SI-HGs) for finding stationary points of nonconvex-nonconcave minimax problems. SI-HGs have the convergence rate O(1/K)O(1/K), which improves upon the rate O(1/K1/2)O(1/K^{1/2}) of DAT and YOPO. We devise a practical variant of SI-HGs, and show that it outperforms other AT algorithms in terms of convergence speed and robustness.

Keywords

Cite

@article{arxiv.2202.10523,
  title  = {Semi-Implicit Hybrid Gradient Methods with Application to Adversarial Robustness},
  author = {Beomsu Kim and Junghoon Seo},
  journal= {arXiv preprint arXiv:2202.10523},
  year   = {2022}
}

Comments

International Conference on Artificial Intelligence and Statistics (AISTATS) 2022

R2 v1 2026-06-24T09:48:41.786Z