AISTATS 2023poster12 citations
Nash Equilibria and Pitfalls of Adversarial Training in Adversarial Robustness Games
Maria-Florina Balcan, Rattana Pukdee, Pradeep Ravikumar, Hongyang Zhang
Abstract
Adversarial training is a standard technique for training adversarially robust models. In this paper, we study adversarial training as an alternating best-response strategy in a 2-player zero-sum game. We prove that even in a simple scenario of a linear classifier and a statistical model that abstracts robust vs. non-robust features, the alternating best response strategy of such game may not converge. On the other hand, a unique pure Nash equilibrium of the game exists and is provably robust. We support our theoretical results with experiments, showing the non-convergence of adversarial training and the robustness of Nash equilibrium.
BibTeX
@InProceedings{pmlr-v206-balcan23a,
title = {Nash Equilibria and Pitfalls of Adversarial Training in Adversarial Robustness Games},
author = {Balcan, Maria-Florina and Pukdee, Rattana and Ravikumar, Pradeep and Zhang, Hongyang},
booktitle = {Proceedings of The 26th International Conference on Artificial Intelligence and Statistics},
pages = {9607--9636},
year = {2023},
editor = {Ruiz, Francisco and Dy, Jennifer and van de Meent, Jan-Willem},
volume = {206},
series = {Proceedings of Machine Learning Research},
month = {25--27 Apr},
publisher = {PMLR},
pdf = {https://proceedings.mlr.press/v206/balcan23a/balcan23a.pdf},
url = {https://proceedings.mlr.press/v206/balcan23a.html},
abstract = {Adversarial training is a standard technique for training adversarially robust models. In this paper, we study adversarial training as an alternating best-response strategy in a 2-player zero-sum game. We prove that even in a simple scenario of a linear classifier and a statistical model that abstracts robust vs. non-robust features, the alternating best response strategy of such game may not converge. On the other hand, a unique pure Nash equilibrium of the game exists and is provably robust. We support our theoretical results with experiments, showing the non-convergence of adversarial training and the robustness of Nash equilibrium.}
}