Accuracy of a ResNeXt-29, 8x64d trained using the full CIFAR-100 dataset (“Baseline”) and two synthetic experiments with altered datasets. (“Subsample”) drops 75% of the dolphin and mountain subclasses from the training dataset, and (“Random Noise”) assigns 25% of examples from these subclasses a random superclass label. Results reported are on superclass labels for the validation set. Numbers in parentheses are reductions in performance with respect to the baseline model for each experimental condition.