Counter-Samples: A Stateless Strategy to Neutralize Black Box Adversarial Attacks

arXiv:2403.10562v1 Announce Type: new
Abstract: Our paper presents a novel defence against black box attacks, where attackers use the victim model as an oracle to craft their adversarial examples. Unlike traditional preprocessing defences that rely on sanitizing input samples, our stateless strategy counters the attack process itself. For every query we evaluate a counter-sample instead, where the counter-sample is the original sample optimized against the attacker's objective. By countering every black box query with a targeted white box optimization, our …

adversarial adversarial attacks arxiv attack attackers attacks black box box counter cs.ai cs.cr cs.lg defence examples input novel oracle process query stateless strategy victim