Ablation-CAM: Visual Explanations for Deep Convolutional Network via Gradient-free Localization

saurabh desai, Harish Guruprasad Ramaswamy; Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2020, pp. 983-991

Abstract


In response to recent criticism of gradient-based visualization techniques, we propose a new methodology to generate visual explanations for deep Convolutional Neural Networks (CNN) - based models. Our approach - Ablation-based Class Activation Mapping (Ablation CAM) uses ablation analysis to determine the importance (weights) of individual feature map units w.r.t. class. Further, this is used to produce a coarse localization map highlighting the important regions in the image for predicting the concept. Our objective and subjective evaluations show that this gradient-free approach works better than state-of-the-art Grad-CAM technique. Moreover, further experiments are carried out to show that Ablation-CAM is class discriminative as well as can be used to evaluate trust in a model.

Related Material


[pdf]
[bibtex]
@InProceedings{desai_2020_WACV,
author = {desai, saurabh and Ramaswamy, Harish Guruprasad},
title = {Ablation-CAM: Visual Explanations for Deep Convolutional Network via Gradient-free Localization},
booktitle = {Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV)},
month = {March},
year = {2020}
}