Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networks
Although deep neural networks (DNNs) have achieved great success in many\ntasks, they can often be fooled by \\emph{adversarial examples} that are\ngenerated by adding small but...