The Odds Are Odd：A Statistical Test for Detecting Adversarial Examples（英文）
The Odds are Odd: A Statistical Test for Detecting Adversarial Examples
摘要：
我们调查的条件下，存在的测试统计数据，可以可靠地检测的例子，这已被恶意操纵的白盒攻击。这些统计数据可以很容易地计算和校准随机损坏的输入。它们利用了对抗攻击引入的某些异常，特别是如果它们遵循在p范数约束下最优选择扰动的范式。访问日志赔率是捍卫模型的唯一要求。我们证明我们的方法经验，但也提供了条件下，通过建议的测试统计的可检测性保证是有效的。在我们的实验中，我们表明，它甚至可以纠正测试时间预测对抗攻击具有高精度。
英文摘要：
We investigate conditions under which test statistics exist that can reliably detect examples, which have been adversarially manipulated in a white-box attack. These statistics can be easily computed and calibrated by randomly corrupting inputs. They exploit certain anomalies that adversarial attacks introduce, in particular if they follow the paradigm of choosing perturbations optimally under p-norm constraints. Access to the log-odds is the only requirement to defend models. We justify our approach empirically, but also provide conditions under which detectability via the suggested test statistics is guaranteed to be effective. In our experiments, we show that it is even possible to correct test time predictions for adversarial attacks with high accuracy.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
