Noisy Feature Mixup
This work addresses the need for more robust models in computer vision, offering an incremental improvement over existing mixup methods.
The paper tackles the problem of improving model robustness and accuracy in computer vision by introducing Noisy Feature Mixup (NFM), a data augmentation method that combines interpolation and noise injection, resulting in better trade-offs between predictive accuracy on clean data and robustness to perturbations across benchmark datasets.
We introduce Noisy Feature Mixup (NFM), an inexpensive yet effective method for data augmentation that combines the best of interpolation based training and noise injection schemes. Rather than training with convex combinations of pairs of examples and their labels, we use noise-perturbed convex combinations of pairs of data points in both input and feature space. This method includes mixup and manifold mixup as special cases, but it has additional advantages, including better smoothing of decision boundaries and enabling improved model robustness. We provide theory to understand this as well as the implicit regularization effects of NFM. Our theory is supported by empirical results, demonstrating the advantage of NFM, as compared to mixup and manifold mixup. We show that residual networks and vision transformers trained with NFM have favorable trade-offs between predictive accuracy on clean data and robustness with respect to various types of data perturbation across a range of computer vision benchmark datasets.