Research

Min-Max Optimization without Gradients: Convergence and Applications to Black-Box Evasion and Poisoning Attacks

ICML

Authors

Published on

07/18/2020

In this paper, we study the problem of constrained min-max optimization in a black-box setting, where the desired optimizer cannot access the gradients of the objective function but may query its values. We present a principled optimization framework, integrating a zeroth-order (ZO) gradient estimator with an alternating projected stochastic gradient descent-ascent method, where the former only requires a small number of function queries and the later needs just one-step descent/ascent update. We show that the proposed framework, referred to as ZO-Min-Max, has a sublinear convergence rate under mild conditions and scales gracefully with problem size. We also explore a promising connection between black-box min-max optimization and black-box evasion and poisoning attacks in adversarial machine learning (ML). Our empirical evaluations on these use cases demonstrate the effectiveness of our approach and its scalability to dimensions that prohibit using recent black-box solvers.

Please cite our work using the BibTeX below.

@InProceedings{pmlr-v119-liu20j,
  title = 	 {Min-Max Optimization without Gradients: Convergence and Applications to Black-Box Evasion and Poisoning Attacks},
  author =       {Liu, Sijia and Lu, Songtao and Chen, Xiangyi and Feng, Yao and Xu, Kaidi and Al-Dujaili, Abdullah and Hong, Mingyi and O'Reilly, Una-May},
  booktitle = 	 {Proceedings of the 37th International Conference on Machine Learning},
  pages = 	 {6282--6293},
  year = 	 {2020},
  editor = 	 {III, Hal Daumé and Singh, Aarti},
  volume = 	 {119},
  series = 	 {Proceedings of Machine Learning Research},
  month = 	 {13--18 Jul},
  publisher =    {PMLR},
  pdf = 	 {http://proceedings.mlr.press/v119/liu20j/liu20j.pdf},
  url = 	 {https://proceedings.mlr.press/v119/liu20j.html}
}
Close Modal