[1] Threat of Adversarial Attacks on Deep Learning in Computer Vision: A Survey. 英文综述 link
[2] 知乎专栏，前几章有介绍样本对抗攻防的基础的 link
[3] pytorch官方文档----60分钟入门 link
[4] 样本对抗的来龙去脉和本质 link
[5] 样本对抗的一个相关比赛 link
[6] Awesome ML Attack link
[7] 简单易懂的人脸识别过程和原理介绍 link
[8] 一种鲁棒的神经网络架构(防御) link
[9] 对抗训练论文一(防御) link
[10] Ian GoodFellow机器学习的博客 link
[1] PyTorch FGSM Tutorial link
[2] PyTorch C&W Attack link
[3] PyTorch DDN Attack(CVPR2019) link
[1] Loss Function for training Face Recognition Model link
[2] Face Recognition Model: ZhaoJ9014/face.evoLVe.PyTorch （默认白盒模型） link
[3] Face Recognition Model: ageitgey/face_recognition （第一次老师给的白盒模型） link
[0] 浙大的一篇调研 link
[1] Neural Cleanse: Identifying and Mitigating Backdoor Attacks in Neural Networks. link 翻译 link
[2] Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoning.(key pattern) link
[3] A General Framework for Adversarial Examples with Objectives.(AGN方法) link 机器之心的解读link
[4] Accessorize to a Crime: Real and Stealthy Attacks on State-of-the-Art Face Recognition. link 源码link
[5] Robust Physical-World Attacks on Deep Learning Visual Classification.(对路牌攻击CVPR2018) link
[0] ZOO: Zeroth Order Optimization based Black-box Attacks to Deep Neural Networks without Training Substitute Models 没有使用替代模型，用零阶替代一阶 link
[1] AutoZOOM: Autoencoder-based Zeroth Order Optimization Method for Attacking Black-box Neural Networks ZOO模型的优化，提高对抗样本的生成效率 link
[2] Delving into Transferable Adversarial Examples and Black-box Attacks 采用集成的方式，用了替代模型求解梯度 link
[3] Ensemble Adversarial Training Attacks and Defenses link
[4] Curls & Whey: Boosting Black-Box Adversarial Attacks(CVPR2019 oral) 主要解决对抗样本的多样性(两种梯度迭代方式)，二分法减小噪声的方差 link 知乎讲解link GitHublink
[5] Black-box Adversarial Attacks with Limited Queries and Information 利用演化算法来攻击 link github源码 (https://github.com/labsix/limited-blackbox-attacks)
[6] SemanticAdv: Generating Adversarial Examples via Attribute-conditional Image Editing 通过人脸编辑来做人脸数据集上的对抗样本 link