Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

HopSkipJumpAttack: A Query-Efficient Decision-Based Attack

About

The goal of a decision-based adversarial attack on a trained model is to generate adversarial examples based solely on observing output labels returned by the targeted model. We develop HopSkipJumpAttack, a family of algorithms based on a novel estimate of the gradient direction using binary information at the decision boundary. The proposed family includes both untargeted and targeted attacks optimized for $\ell_2$ and $\ell_\infty$ similarity metrics respectively. Theoretical analysis is provided for the proposed algorithms and the gradient direction estimate. Experiments show HopSkipJumpAttack requires significantly fewer model queries than Boundary Attack. It also achieves competitive performance in attacking several widely-used defense mechanisms. (HopSkipJumpAttack was named Boundary Attack++ in a previous version of the preprint.)

Jianbo Chen, Michael I. Jordan, Martin J. Wainwright• 2019

Related benchmarks

TaskDatasetResultRank
Object Hallucination EvaluationPOPE--
2019
Black-box AttackLSUN
ASR95.8
189
Black-box AttackGenImage
ASR99.2
162
Targeted Adversarial AttackImageNet ILSVRC2012 (val)
Median L2 Distance78.626
120
Targeted Adversarial AttackImageNet (val)
Median L2 Distance76.727
120
Adversarial AttackILSVRC 2012 (val)
Median L2 Distance24.181
112
Targeted Black-box Adversarial AttackImageNet
Average L2 Norm10.715
96
Non-targeted Black-box Adversarial AttackImageNet
Average L2 Norm2.55
96
Adversarial AttackILSVRC 2012
Median L2 Distance17.75
96
Adversarial AttackImageNet-21K (val)
Median L2 Distance4.367
80
Showing 10 of 42 rows

Other info

Follow for update