Quantification of Uncertainty with Adversarial Models