Visualizing Deep Networks by Optimizing with Integrated Gradients

Zhongang Qi, Saeed Khorram, Fuxin Li; Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops, 2019, pp. 1-4

Abstract


Understanding and interpreting the decisions made by deep learning models is valuable in many domains. In computer vision, computing heatmaps from a deep network is a popular approach for visualizing and understanding deep networks. However, heatmaps that do not correlate with the network may mislead human, hence the performance of heatmaps in providing a faithful explanation to the underlying deep network is crucial. In this paper, we propose I-GOS, which optimizes for a heatmap so that the classification scores on the masked image would maximally decrease. The main novelty of the approach is to compute descent directions based on the integrated gradients instead of the normal gradient, which avoids local optima and speeds up convergence. Extensive experiments show that the heatmaps produced by our approach are more correlated with the decision of the underlying deep network, in comparison with other state-of-the-art approaches.

Related Material


[pdf] [dataset]
[bibtex]
@InProceedings{Qi_2019_CVPR_Workshops,
author = {Qi, Zhongang and Khorram, Saeed and Li, Fuxin},
title = {Visualizing Deep Networks by Optimizing with Integrated Gradients},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2019}
}