Adversarial examples are inputs to machine learning models that an attacker has intentionally designed to confuse the model into making a mistake. Such examples pose a serious threat to the applicability of machine-learning-based systems, especially in life- and safety-critical domains. To address this problem, the area of adversarial robustness investigates mechanisms behind adversarial attacks and defenses against these attacks. This survey reviews literature that focuses on the effects of data used by a model on the model's adversarial robustness. It systematically identifies and summarizes the state-of-the-art research in this area and further discusses gaps of knowledge and promising future research directions.

翻译：对抗样本是攻击者故意设计的、旨在使机器学习模型产生错误的输入。此类样本对基于机器学习的系统（尤其是在生命安全和关键安全领域）的适用性构成了严重威胁。为解决这一问题，对抗鲁棒性领域研究了对抗攻击背后的机制及针对这些攻击的防御方法。本综述回顾了聚焦于模型所用数据对模型对抗鲁棒性影响的相关文献，系统性地识别并总结了该领域的最新研究进展，进一步讨论了知识空白与有前景的未来研究方向。