On Robustness and Transferability of Convolutional Neural Networks
This work addresses robustness challenges in CNNs for image classification, providing insights for improving generalization under distribution shifts, though it is incremental in nature.
The paper investigates how out-of-distribution robustness and transfer performance in CNNs are affected by factors like pre-training data size, model scale, and data preprocessing, finding that larger training sets and models improve robustness, and simple preprocessing changes can mitigate issues.
Modern deep convolutional networks (CNNs) are often criticized for not generalizing under distributional shifts. However, several recent breakthroughs in transfer learning suggest that these networks can cope with severe distribution shifts and successfully adapt to new tasks from a few training examples. In this work we study the interplay between out-of-distribution and transfer performance of modern image classification CNNs for the first time and investigate the impact of the pre-training data size, the model scale, and the data preprocessing pipeline. We find that increasing both the training set and model sizes significantly improve the distributional shift robustness. Furthermore, we show that, perhaps surprisingly, simple changes in the preprocessing such as modifying the image resolution can significantly mitigate robustness issues in some cases. Finally, we outline the shortcomings of existing robustness evaluation datasets and introduce a synthetic dataset SI-Score we use for a systematic analysis across factors of variation common in visual data such as object size and position.