IDGI: A Framework to Eliminate Explanation Noise from Integrated Gradients

Abstract

Integrated Gradients (IG) as well as its variants are well-known techniques for interpreting the decisions of deep neural networks. While IG-based approaches attain state-of-the-art performance, they often integrate noise into their explanation saliency maps, which reduce their interpretability. To minimize the noise, we examine the source of the noise analytically and propose a new approach to reduce the explanation noise based on our analytical findings. We propose the Important Direction Gradient Integration (IDGI) framework, which can be easily incorporated into any IG-based method that uses the Reimann Integration for integrated gradient computation. Extensive experiments with three IG-based methods show that IDGI improves them drastically on numerous interpretability metrics.

Cite

Text

Yang et al. "IDGI: A Framework to Eliminate Explanation Noise from Integrated Gradients." Conference on Computer Vision and Pattern Recognition, 2023. doi:10.1109/CVPR52729.2023.02272

Markdown

[Yang et al. "IDGI: A Framework to Eliminate Explanation Noise from Integrated Gradients." Conference on Computer Vision and Pattern Recognition, 2023.](https://mlanthology.org/cvpr/2023/yang2023cvpr-idgi/) doi:10.1109/CVPR52729.2023.02272

BibTeX

@inproceedings{yang2023cvpr-idgi,
  title     = {{IDGI: A Framework to Eliminate Explanation Noise from Integrated Gradients}},
  author    = {Yang, Ruo and Wang, Binghui and Bilgic, Mustafa},
  booktitle = {Conference on Computer Vision and Pattern Recognition},
  year      = {2023},
  pages     = {23725-23734},
  doi       = {10.1109/CVPR52729.2023.02272},
  url       = {https://mlanthology.org/cvpr/2023/yang2023cvpr-idgi/}
}