Evading Black-box Classifiers Without Breaking Eggs

arXiv:2306.02895v2 Announce Type: replace
Abstract: Decision-based evasion attacks repeatedly query a black-box classifier to generate adversarial examples. Prior work measures the cost of such attacks by the total number of queries made to the classifier. We argue this metric is flawed. Most security-critical machine learning systems aim to weed out "bad" data (e.g., malware, harmful content, etc). Queries to such systems carry a fundamentally asymmetric cost: queries detected as "bad" come at a higher cost because they trigger additional security …

adversarial aim arxiv attacks bad box breaking cost critical cs.cr cs.lg data decision evasion evasion attacks examples machine machine learning malware metric query security stat.ml systems work