Adversarial Training for Gradient Descent: Analysis Through its Continuous-time Approximation
Gu, Haotian, Guo, Xin, Li, Xinyu
–arXiv.org Artificial Intelligence
Adversarial training has gained great popularity as one of the most effective defenses for deep neural network and more generally for gradient-based machine learning models against adversarial perturbations on data points. This paper establishes a continuous-time approximation for the mini-max game of adversarial training. This approximation approach allows for precise and analytical comparisons between stochastic gradient descent and its adversarial training counterpart; and confirms theoretically the robustness of adversarial training from a new gradient-flow viewpoint. The analysis is then corroborated through various analytical and numerical examples.
arXiv.org Artificial Intelligence
May-24-2023