A simple way to make neural networks robust against diverse image corruptions

• Main Authors: Rusak, Evgenia, Schott, Lukas, Zimmermann, Roland S, Bitterwolf, Julian, Bringmann, Oliver, Bethge, Matthias, Brendel, Wieland
• Format: Journal Article
• Language: English
• Published: 16-01-2020
• Subjects: Computer Science - Computer Vision and Pattern Recognition; Computer Science - Learning; Statistics - Machine Learning
• DOI: 10.48550/arxiv.2001.06057

The human visual system is remarkably robust against a wide range of
naturally occurring variations and corruptions like rain or snow. In contrast,
the performance of modern image recognition models strongly degrades when
evaluated on previously unseen corruptions. Here, we demonstrate that a simple
but properly tuned training with additive Gaussian and Speckle noise
generalizes surprisingly well to unseen corruptions, easily reaching the
previous state of the art on the corruption benchmark ImageNet-C (with
ResNet50) and on MNIST-C. We build on top of these strong baseline results and
show that an adversarial training of the recognition model against uncorrelated
worst-case noise distributions leads to an additional increase in performance.
This regularization can be combined with previously proposed defense methods
for further improvement.