PRIME: A few primitives can boost robustness to common corruptions

TL;DR

PRIME introduces a simple, principled data augmentation method using max-entropy image transformations that significantly improves robustness to common corruptions in image classification tasks, while being easy to implement and combine with other techniques.

Contribution

It presents PRIME, a novel, simple data augmentation scheme based on max-entropy transformations that enhances corruption robustness and is computationally efficient.

Findings

PRIME outperforms prior methods in corruption robustness.

The mixing strategy is crucial for synthesizing corrupted images.

PRIME's efficiency allows easy integration into various training schemes.

Abstract

Despite their impressive performance on image classification tasks, deep networks have a hard time generalizing to unforeseen corruptions of their data. To fix this vulnerability, prior works have built complex data augmentation strategies, combining multiple methods to enrich the training data. However, introducing intricate design choices or heuristics makes it hard to understand which elements of these methods are indeed crucial for improving robustness. In this work, we take a step back and follow a principled approach to achieve robustness to common corruptions. We propose PRIME, a general data augmentation scheme that relies on simple yet rich families of max-entropy image transformations. PRIME outperforms the prior art in terms of corruption robustness, while its simplicity and plug-and-play nature enable combination with other methods to further boost their robustness. We analyze PRIME to shed light on the importance of the mixing strategy on synthesizing corrupted images, and to reveal the robustness-accuracy trade-offs arising in the context of common corruptions. Finally, we show that the computational efficiency of our method allows it to be easily used in both on-line and off-line data augmentation schemes.