Clarifying Myths About the Relationship Between Shape Bias, Accuracy, and Robustness

TL;DR

This paper critically examines the belief that increasing shape bias in models enhances their robustness to out-of-distribution data, revealing that the relationship is not straightforward and depends on the type of data augmentation used.

Contribution

The study systematically evaluates 39 data augmentations, debunks the shape bias-robustness myth, and highlights dataset biases and the potential to improve both accuracy and robustness simultaneously.

Findings

Shape bias increase does not guarantee higher OOD robustness.

Proper data augmentation can reduce dataset biases.

In-domain accuracy and OOD robustness can be improved together.

Abstract

Deep learning models can perform well when evaluated on images from the same distribution as the training set. However, applying small perturbations in the forms of noise, artifacts, occlusions, blurring, etc. to a model's input image and feeding the model with out-of-distribution (OOD) data can significantly drop the model's accuracy, making it not applicable to real-world scenarios. Data augmentation is one of the well-practiced methods to improve model robustness against OOD data; however, examining which augmentation type to choose and how it affects the OOD robustness remains understudied. There is a growing belief that augmenting datasets using data augmentations that improve a model's bias to shape-based features rather than texture-based features results in increased OOD robustness for Convolutional Neural Networks trained on the ImageNet-1K dataset. This is usually stated as ``an increase in the model's shape bias results in an increase in its OOD robustness". Based on this hypothesis, some works in the literature aim to find augmentations with higher effects on model shape bias and use those for data augmentation. By evaluating 39 types of data augmentations on a widely used OOD dataset, we demonstrate the impact of each data augmentation on the model's robustness to OOD data and further show that the mentioned hypothesis is not true; an increase in shape bias does not necessarily result in higher OOD robustness. By analyzing the results, we also find some biases in the ImageNet-1K dataset that can easily be reduced using proper data augmentation. Our evaluation results further show that there is not necessarily a trade-off between in-domain accuracy and OOD robustness, and choosing the proper augmentations can help increase both in-domain accuracy and OOD robustness simultaneously.