Linked Presentation: Adversarial Preprocessing: Understanding and Preventing Image-Scaling Attacks in Machine Learning