Toward Practical Adversarial Attacks on Face Verification Systems
Toward Practical Adversarial Attacks on Face Verification Systems
复制标题
DOI：
10.1109/biosig52210.2021.9548310
复制
发表时间：
2021-09
期刊：
影响因子：
--
通讯作者：
Kazuya Kakizaki;Taiki Miyagawa;Inderjeet Singh;Jun Sakuma
中科院分区：
文献类型：
--
作者：
Kazuya Kakizaki;Taiki Miyagawa;Inderjeet Singh;Jun Sakuma
基于DNN的人脸验证系统容易受到对抗性样本的攻击。上一篇论文的评估协议（场景），我们称之为探针依赖攻击场景，是不现实的。我们定义了一个更实际的攻击场景，探针不可知攻击。我们的经验表明，这些攻击比探针依赖的更具挑战性。我们提出了一个简单而有效的方法，PAMTAM，以提高攻击成功率的探针不可知攻击。我们表明，PAMTAM成功地提高了攻击成功率在各种各样的实验设置。
DNN-based face verification systems are vulnerable to adversarial examples. The previous paper’s evaluation protocol (scenario), which we called the probe-dependent attack scenario, was unrealistic. We define a more practical attack scenario, the probe-agnostic attack. We empirically show that these attacks are more challenging than probe-dependent ones. We propose a simple and effective method, PAMTAM, to improve the attack success rate for probe-agnostic attacks. We show that PAMTAM successfully improves the attack success rate in a wide variety of experimental settings.