AR2: Attention-Guided Repair for the Robustness of CNNs Against Common Corruptions

TL;DR

AR2 is a simple, iterative method that improves CNNs' robustness to common corruptions by aligning class activation maps between clean and corrupted images, without changing the model architecture.

Contribution

AR2 introduces an attention-guided repair strategy that enhances corruption robustness of pretrained CNNs through CAM alignment and iterative fine-tuning.

Findings

AR2 outperforms existing methods on CIFAR-10-C, CIFAR-100-C, and ImageNet-C.

It maintains high accuracy on clean data while improving robustness.

The method is scalable and does not require architectural modifications.

Abstract

Deep neural networks suffer from significant performance degradation when exposed to common corruptions such as noise, blur, weather, and digital distortions, limiting their reliability in real-world applications. In this paper, we propose AR2 (Attention-Guided Repair for Robustness), a simple yet effective method to enhance the corruption robustness of pretrained CNNs. AR2 operates by explicitly aligning the class activation maps (CAMs) between clean and corrupted images, encouraging the model to maintain consistent attention even under input perturbations. Our approach follows an iterative repair strategy that alternates between CAM-guided refinement and standard fine-tuning, without requiring architectural changes. Extensive experiments show that AR2 consistently outperforms existing state-of-the-art methods in restoring robustness on standard corruption benchmarks (CIFAR-10-C, CIFAR-100-C and ImageNet-C), achieving a favorable balance between accuracy on clean data and corruption robustness. These results demonstrate that AR2 provides a robust and scalable solution for enhancing model reliability in real-world environments with diverse corruptions.