Fourier-Based Augmentations for Improved Robustness and Uncertainty Calibration
This work addresses robustness in computer vision models for applications requiring tailored defenses against specific corruptions, representing an incremental improvement over existing augmentation methods.
The paper tackled the problem of improving model robustness against specific corruptions without sacrificing overall performance, achieving over ten percentage point reductions in classification error for high-severity noise and digital-type corruptions on CIFAR-10-C and CIFAR-100-C datasets.
Diverse data augmentation strategies are a natural approach to improving robustness in computer vision models against unforeseen shifts in data distribution. However, the ability to tailor such strategies to inoculate a model against specific classes of corruptions or attacks -- without incurring substantial losses in robustness against other classes of corruptions -- remains elusive. In this work, we successfully harden a model against Fourier-based attacks, while producing superior-to-AugMix accuracy and calibration results on both the CIFAR-10-C and CIFAR-100-C datasets; classification error is reduced by over ten percentage points for some high-severity noise and digital-type corruptions. We achieve this by incorporating Fourier-basis perturbations in the AugMix image-augmentation framework. Thus we demonstrate that the AugMix framework can be tailored to effectively target particular distribution shifts, while boosting overall model robustness.