Testing robustness of predictions of trained classifiers against naturally occurring perturbations

Open in new window