Hard-label based Small Query Black-box Adversarial Attack

arXiv:2403.06014v1 Announce Type: new
Abstract: We consider the hard label based black box adversarial attack setting which solely observes predicted classes from the target model. Most of the attack methods in this setting suffer from impractical number of queries required to achieve a successful attack. One approach to tackle this drawback is utilising the adversarial transferability between white box surrogate models and black box target model. However, the majority of the methods adopting this approach are soft label based to …

abstract adversarial arxiv attack methods black box box cs.ai cs.cv cs.lg queries query small type