On the Impact of Interpretability Methods in Active Image Augmentation Method

TL;DR

This paper investigates how different interpretability methods affect the effectiveness of Activate Image Augmentation (ADA) in improving the robustness of deep learning models in computer vision, especially under noisy or occluded conditions.

Contribution

The study provides a comprehensive experimental analysis of five interpretability methods within ADA, highlighting GradCam's superior impact on convergence speed.

Findings

All interpretability methods yield similar final performance.

GradCam combined with ADA accelerates model convergence.

Interpretability methods influence training dynamics, not just final accuracy.

Abstract

Robustness is a significant constraint in machine learning models. The performance of the algorithms must not deteriorate when training and testing with slightly different data. Deep neural network models achieve awe-inspiring results in a wide range of applications of computer vision. Still, in the presence of noise or region occlusion, some models exhibit inaccurate performance even with data handled in training. Besides, some experiments suggest deep learning models sometimes use incorrect parts of the input information to perform inference. Activate Image Augmentation (ADA) is an augmentation method that uses interpretability methods to augment the training data and improve its robustness to face the described problems. Although ADA presented interesting results, its original version only used the Vanilla Backpropagation interpretability to train the U-Net model. In this work, we propose an extensive experimental analysis of the interpretability method's impact on ADA. We use five interpretability methods: Vanilla Backpropagation, Guided Backpropagation, GradCam, Guided GradCam, and InputXGradient. The results show that all methods achieve similar performance at the ending of training, but when combining ADA with GradCam, the U-Net model presented an impressive fast convergence.