Questions for the paper: "Same Same But DifferNet: Semi-Supervised Defect Detection with Normalizing Flows" ----------------------------------------------------------------------------------------------- Please send your answers to: galessos@cs.uni-freiburg.de by 10:00 on 04.02.2020 1) What distinguishes normalizing flows from the other types of deep generative models discussed in the paper? How can this be used for anomaly detection? (3 sentences) 2) How does DifferNet ensure robustness in the detection of anomalies? (2 sentences) 3) What method ranks second best in most categories on the MVTec AD dataset? Why do you think this is the case? (2-3 sentences)