ML Anthology

Visualizing the Impact of Feature Attribution Baselines

Pascal Sturmfels, Scott Lundberg, Su-In Lee
Distill 2020
doi:10.23915/distill.00022 /distill/2020/sturmfels2020distill-visualizing/

Abstract

Distill articles are interactive publications and do not include traditional abstracts. This summary was written for the ML Anthology. Investigates how the choice of baseline input in integrated gradients affects feature attribution results, demonstrating that different baselines encode different assumptions about feature missingness and can significantly alter interpretations.

Distill Code Semantic Scholar

Cite

Text

Sturmfels et al. "Visualizing the Impact of Feature Attribution Baselines." Distill, 2020. doi:10.23915/distill.00022

Markdown

[Sturmfels et al. "Visualizing the Impact of Feature Attribution Baselines." Distill, 2020.](https://mlanthology.org/distill/2020/sturmfels2020distill-visualizing/) doi:10.23915/distill.00022

BibTeX

@article{sturmfels2020distill-visualizing,
  title     = {{Visualizing the Impact of Feature Attribution Baselines}},
  author    = {Sturmfels, Pascal and Lundberg, Scott and Lee, Su-In},
  journal   = {Distill},
  year      = {2020},
  doi       = {10.23915/distill.00022},
  url       = {https://mlanthology.org/distill/2020/sturmfels2020distill-visualizing/}
}