Beware the Black-Box: On the Robustness of Recent Defenses to Adversarial Examples

• Published in: Entropy (Basel, Switzerland) Vol. 23; no. 10; p. 1359
• Main Authors: Mahmood, Kaleel, Gurevin, Deniz, van Dijk, Marten, Nguyen, Phuoung Ha
• Format: Journal Article
• Language: English
• Published: Basel MDPI AG 18.10.2021; MDPI
• Subjects: adversarial machine learning; black-box attacks; Buffer zones; Distillation; Error correcting codes; Error correction; Experiments; Machine learning; Neural networks; security
• ISSN: 1099-4300; 1099-4300
• DOI: 10.3390/e23101359

Many defenses have recently been proposed at venues like NIPS, ICML, ICLR and CVPR. These defenses are mainly focused on mitigating white-box attacks. They do not properly examine black-box attacks. In this paper, we expand upon the analyses of these defenses to include adaptive black-box adversaries. Our evaluation is done on nine defenses including Barrage of Random Transforms, ComDefend, Ensemble Diversity, Feature Distillation, The Odds are Odd, Error Correcting Codes, Distribution Classifier Defense, K-Winner Take All and Buffer Zones. Our investigation is done using two black-box adversarial models and six widely studied adversarial attacks for CIFAR-10 and Fashion-MNIST datasets. Our analyses show most recent defenses (7 out of 9) provide only marginal improvements in security (<25%), as compared to undefended networks. For every defense, we also show the relationship between the amount of data the adversary has at their disposal, and the effectiveness of adaptive black-box attacks. Overall, our results paint a clear picture: defenses need both thorough white-box and black-box analyses to be considered secure. We provide this large scale study and analyses to motivate the field to move towards the development of more robust black-box defenses.