AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Explaining Neural Networks Semantically and Quantitatively

Runjin Chen, Hao Chen, Jie Ren, Ge Huang, Quanshi Zhang

ICCV 2019 · Label-efficient and robust learning › Recognition and long-tail · approach: Self-explaining and visual explanation networks · 59 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add prior-weight losses to the distillation-only explainerexplainer, AlexNet performer · Pascal-Part animal classificationentropy of contributions: 3.840 -> 5.136Improved
add prior-weight losses to the distillation-only explainerexplainer, VGG-S performer · Pascal-Part animal classificationentropy of contributions: 4.078 -> 5.831Improved
add prior-weight losses to the distillation-only explainerexplainer, VGG-16 performer · Pascal-Part animal classificationentropy of contributions: 4.392 -> 5.932Improved
add prior-weight losses to the distillation-only explainerexplainer, VGG-16 face attribute CNN · CelebA face attributesentropy of contributions: 3.142 -> 3.201Improved
add prior-weight losses to the distillation-only explainerexplainer · Pascal-Part object partserror of estimated part contribution: 0.164 -> 0.089Improved
add prior-weight losses to the distillation-only explainerexplainer · Pascal-Part eye parterror of estimated part contribution: 0.399 -> 0.181Improved
replace the performer by the explainer for predictionVGG-16 face attribute CNN · CelebA face attributesclassification accuracy: 90.2 -> 85.9Hurt
replace the performer by the explainer for predictionAlexNet · Pascal-Part animal classificationclassification accuracy: 93.9 -> 92.6Hurt
replace the performer by the explainer for predictionVGG-16 · Pascal-Part animal classificationclassification accuracy: 95.4 -> 95.6No clear effect
use more visual concepts for explanation (ablation)explainer · CelebA face attributesexplanation capacityImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy