Jihoon Tack, Sihyun Yu, Jongheon Jeong, Minseon Kim, Sung Ju Hwang...
TL;DR本文提出了一种通过优化辅助一致性规则损失来避免鲁棒过度拟合的有效正则化技术，在 Adversarial training 过程中使用数据扩增来强制攻击后的预测分布相似。实验结果表明，这种简单的方法可以显著提高各种 AT 方法的测试准确性，并对模型作出更具实际意义的泛化。
Abstract
adversarial training (AT) is currently one of the most successful methods to obtain the adversarial robustness of deep neural networks. However, a significant generalization gap in the robustness obtained from AT has been problematic, making practitioners to consider a bag of tricks fo