Frequency-Based Vulnerability Analysis of Deep Learning Models against Image Corruptions
This addresses the vulnerability of deep learning models to diverse image corruptions for applications like autonomous driving or security, but it is incremental as it builds on existing corruption analysis.
The paper tackled the problem of deep learning models failing on real-world image corruptions not covered by existing datasets, and found that even robust models struggle against low visibility-based corruptions crafted by their MUFIA algorithm.
Deep learning models often face challenges when handling real-world image corruptions. In response, researchers have developed image corruption datasets to evaluate the performance of deep neural networks in handling such corruptions. However, these datasets have a significant limitation: they do not account for all corruptions encountered in real-life scenarios. To address this gap, we present MUFIA (Multiplicative Filter Attack), an algorithm designed to identify the specific types of corruptions that can cause models to fail. Our algorithm identifies the combination of image frequency components that render a model susceptible to misclassification while preserving the semantic similarity to the original image. We find that even state-of-the-art models trained to be robust against known common corruptions struggle against the low visibility-based corruptions crafted by MUFIA. This highlights the need for more comprehensive approaches to enhance model robustness against a wider range of real-world image corruptions.