Improving Robustness to Out-of-Distribution Data by Frequency-based Augmentation

TL;DR

This paper introduces a frequency-based data augmentation method that enhances CNN robustness to out-of-distribution data by replacing frequency components, significantly improving detection performance on CIFAR10 versus SVHN.

Contribution

The paper proposes a novel frequency-based augmentation technique that improves CNN robustness to out-of-distribution data, demonstrated by increased AUROC scores on CIFAR10 and SVHN datasets.

Findings

AUROC increased from 89.22% to 98.15% with the proposed method

Combining with other augmentations further improves AUROC to 98.59%

Robust models rely heavily on high-frequency image components

Abstract

Although Convolutional Neural Networks (CNNs) have high accuracy in image recognition, they are vulnerable to adversarial examples and out-of-distribution data, and the difference from human recognition has been pointed out. In order to improve the robustness against out-of-distribution data, we present a frequency-based data augmentation technique that replaces the frequency components with other images of the same class. When the training data are CIFAR10 and the out-of-distribution data are SVHN, the Area Under Receiver Operating Characteristic (AUROC) curve of the model trained with the proposed method increases from 89.22\% to 98.15\%, and further increased to 98.59\% when combined with another data augmentation method. Furthermore, we experimentally demonstrate that the robust model for out-of-distribution data uses a lot of high-frequency components of the image.