Non-adversarial Robustness of Deep Learning Methods for Computer Vision

TL;DR

This paper reviews recent techniques and benchmark datasets for improving and evaluating the natural robustness of deep learning models in computer vision against distribution shifts caused by natural variations.

Contribution

It provides a comprehensive overview of current methods, datasets, and trends in enhancing deep learning robustness for computer vision applications.

Findings

Summarizes recent approaches to robustness in computer vision.

Reviews benchmark datasets for evaluating natural robustness.

Identifies strengths, limitations, and trends in robustness methods.

Abstract

Non-adversarial robustness, also known as natural robustness, is a property of deep learning models that enables them to maintain performance even when faced with distribution shifts caused by natural variations in data. However, achieving this property is challenging because it is difficult to predict in advance the types of distribution shifts that may occur. To address this challenge, researchers have proposed various approaches, some of which anticipate potential distribution shifts, while others utilize knowledge about the shifts that have already occurred to enhance model generalizability. In this paper, we present a brief overview of the most recent techniques for improving the robustness of computer vision methods, as well as a summary of commonly used robustness benchmark datasets for evaluating the model's performance under data distribution shifts. Finally, we examine the strengths and limitations of the approaches reviewed and identify general trends in deep learning robustness improvement for computer vision.