Keywords: adversarial example defense; deep neural network; robust machine learning; safe machine learning; trust.