This is the code used in the paper "Utilizing Class Separation Distance for the Evaluation of Corruption Robustness of Machine Learning Classifiers" published at 2022 AISafety Workshop at IJCAI in Vienna. Full text can be found here: https://arxiv.org/abs/2206.13405

For image data, experiments in the paper have been carried out with L-infinity norm data augmentations, while this code now provide the possibility to augment L2 and Gaussian noise as well.