Parsimonious Black-Box Adversarial Attacks via Efficient Combinatorial Optimization

Seungyong Moon, Gaon An, Hyun Oh Song
Proceedings of the 36th International Conference on Machine Learning, PMLR 97:4636-4645, 2019.

Abstract

Solving for adversarial examples with projected gradient descent has been demonstrated to be highly effective in fooling the neural network based classifiers. However, in the black-box setting, the attacker is limited only to the query access to the network and solving for a successful adversarial example becomes much more difficult. To this end, recent methods aim at estimating the true gradient signal based on the input queries but at the cost of excessive queries. We propose an efficient discrete surrogate to the optimization problem which does not require estimating the gradient and consequently becomes free of the first order update hyperparameters to tune. Our experiments on Cifar-10 and ImageNet show the state of the art black-box attack performance with significant reduction in the required queries compared to a number of recently proposed methods. The source code is available at https://github.com/snu-mllab/parsimonious-blackbox-attack.

Cite this Paper

BibTeX
@InProceedings{pmlr-v97-moon19a, title = {Parsimonious Black-Box Adversarial Attacks via Efficient Combinatorial Optimization}, author = {Moon, Seungyong and An, Gaon and Song, Hyun Oh}, booktitle = {Proceedings of the 36th International Conference on Machine Learning}, pages = {4636--4645}, year = {2019}, editor = {Chaudhuri, Kamalika and Salakhutdinov, Ruslan}, volume = {97}, series = {Proceedings of Machine Learning Research}, month = {09--15 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v97/moon19a/moon19a.pdf}, url = {https://proceedings.mlr.press/v97/moon19a.html}, abstract = {Solving for adversarial examples with projected gradient descent has been demonstrated to be highly effective in fooling the neural network based classifiers. However, in the black-box setting, the attacker is limited only to the query access to the network and solving for a successful adversarial example becomes much more difficult. To this end, recent methods aim at estimating the true gradient signal based on the input queries but at the cost of excessive queries. We propose an efficient discrete surrogate to the optimization problem which does not require estimating the gradient and consequently becomes free of the first order update hyperparameters to tune. Our experiments on Cifar-10 and ImageNet show the state of the art black-box attack performance with significant reduction in the required queries compared to a number of recently proposed methods. The source code is available at https://github.com/snu-mllab/parsimonious-blackbox-attack.} }
Endnote
%0 Conference Paper %T Parsimonious Black-Box Adversarial Attacks via Efficient Combinatorial Optimization %A Seungyong Moon %A Gaon An %A Hyun Oh Song %B Proceedings of the 36th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2019 %E Kamalika Chaudhuri %E Ruslan Salakhutdinov %F pmlr-v97-moon19a %I PMLR %P 4636--4645 %U https://proceedings.mlr.press/v97/moon19a.html %V 97 %X Solving for adversarial examples with projected gradient descent has been demonstrated to be highly effective in fooling the neural network based classifiers. However, in the black-box setting, the attacker is limited only to the query access to the network and solving for a successful adversarial example becomes much more difficult. To this end, recent methods aim at estimating the true gradient signal based on the input queries but at the cost of excessive queries. We propose an efficient discrete surrogate to the optimization problem which does not require estimating the gradient and consequently becomes free of the first order update hyperparameters to tune. Our experiments on Cifar-10 and ImageNet show the state of the art black-box attack performance with significant reduction in the required queries compared to a number of recently proposed methods. The source code is available at https://github.com/snu-mllab/parsimonious-blackbox-attack.
APA
Moon, S., An, G. & Song, H.O.. (2019). Parsimonious Black-Box Adversarial Attacks via Efficient Combinatorial Optimization. Proceedings of the 36th International Conference on Machine Learning, in Proceedings of Machine Learning Research 97:4636-4645 Available from https://proceedings.mlr.press/v97/moon19a.html.

Related Material