We present CartoonX (Cartoon Explanation), a novel model-agnostic explanation method tailored towards image classifiers and based on the rate-distortion explanation (RDE) framework. Natural images are roughly piece-wise smooth signals -- also called cartoon-like images -- and tend to be sparse in the wavelet domain. CartoonX is the first explanation method to exploit this by requiring its explanations to be sparse in the wavelet domain, thus extracting the relevant piece-wise smooth part of an image instead of relevant pixel-sparse regions. We demonstrate that CartoonX can reveal novel valuable explanatory information, particularly for misclassifications. Moreover, we show that CartoonX achieves a lower distortion with fewer coefficients than other state-of-the-art methods.
Cartoon Explanations of Image Classifiers
CartoonX is a model-agnostic method for image classification explanations that leverages wavelet sparsity to reveal meaningful, piece-wise smooth regions in images with less distortion than other methods.
- Year
- 2021
- Venue
- cartoon-explanations-of-image-classifiers-1
- Authors
- 5
- Hosting
- Abstract onlyARXIV-DEFAULT
Cite
Notes
Only stored in your browser.
Attribution
- Abstract & full text
- arxiv.org/abs/2110.03485v5ARXIV-DEFAULT
- TL;DR
- Semantic Scholar