Top-GAP: Integrating Size Priors in CNNs for more Interpretability, Robustness, and Bias Mitigation

TL;DR

Top-GAP is a regularization method for CNNs that improves interpretability, robustness, and bias mitigation by constraining feature size, leading to better focus on salient regions and enhanced performance against attacks and shifts.

Contribution

The paper introduces Top-GAP, a novel size prior regularization technique that enhances CNN interpretability and robustness without sacrificing clean accuracy.

Findings

Achieves over 50% robust accuracy on CIFAR-10 with PGD attack.

Up to 25% improvement in object localization IOU.

Increases accuracy against distribution shifts by up to 5%.

Abstract

This paper introduces Top-GAP, a novel regularization technique that enhances the explainability and robustness of convolutional neural networks. By constraining the spatial size of the learned feature representation, our method forces the network to focus on the most salient image regions, effectively reducing background influence. Using adversarial attacks and the Effective Receptive Field, we show that Top-GAP directs more attention towards object pixels rather than the background. This leads to enhanced interpretability and robustness. We achieve over 50% robust accuracy on CIFAR-10 with PGD $\epsilon=\frac{8}{255}$ and $20$ iterations while maintaining the original clean accuracy. Furthermore, we see increases of up to 5% accuracy against distribution shifts. Our approach also yields more precise object localization, as evidenced by up to 25% improvement in Intersection over Union (IOU) compared to methods like GradCAM and Recipro-CAM.