StackMix: A complementary Mix algorithm
This work provides an incremental improvement in data augmentation techniques for training convolutional neural networks, benefiting researchers and practitioners seeking to enhance model performance and robustness.
This paper introduces StackMix, a data augmentation technique that concatenates two images as input and averages their one-hot labels. StackMix alone performs comparably to other Mix methods, but its main contribution is achieving significant gains when combined with existing Mix augmentations, such as improving ImageNet test error by 0.8% when combined with CutMix and CIFAR-100-C robustness by 0.7% when combined with AugMix.
Techniques combining multiple images as input/output have proven to be effective data augmentations for training convolutional neural networks. In this paper, we present StackMix: Each input is presented as a concatenation of two images, and the label is the mean of the two one-hot labels. On its own, StackMix rivals other widely used methods in the "Mix" line of work. More importantly, unlike previous work, significant gains across a variety of benchmarks are achieved by combining StackMix with existing Mix augmentation, effectively mixing more than two images. E.g., by combining StackMix with CutMix, test error in the supervised setting is improved across a variety of settings over CutMix, including 0.8\% on ImageNet, 3\% on Tiny ImageNet, 2\% on CIFAR-100, 0.5\% on CIFAR-10, and 1.5\% on STL-10. Similar results are achieved with Mixup.We further show that gains hold for robustness to common input corruptions and perturbations at varying severities with a 0.7\% improvement on CIFAR-100-C, by combining StackMix with AugMix over AugMix. On its own, improvements with StackMix hold across different number of labeled samples on CIFAR-100, maintaining approximately a 2\% gap in test accuracy -- down to using only 5\% of the whole dataset -- and is effective in the semi-supervised setting with a 2\% improvement with the standard benchmark $Π$-model. Finally, we perform an extensive ablation study to better understand the proposed algorithm.