AI Paper Atlas

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

IG 2 : Integrated Gradient on Iterative Gradient Path for Feature Attribution

Yue Zhuo, Zhiqiang Ge

TPAMI 2024 · Label-efficient and robust learning › Recognition and long-tail · approach: Feature attribution methods · 47 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 GradCF baseline instead of black baseline on straight-line path (ablation)Vanilla IG · WM-811K wafer mapground truth SUM: 0.342 -> 0.510Improved
add GradCF baseline instead of train data baseline on straight-line path (ablation)Vanilla IG · WM-811K wafer mapSIC AUC-ADD: 0.777 -> 0.842Improved
add GradCF baseline instead of black baseline on Guided IG path (ablation)Guided IG · WM-811K wafer mapSIC AUC-DEL: 0.050 -> 0.037Improved
add GradCF baseline instead of train data baseline on Guided IG path (ablation)Guided IG · WM-811K wafer mapground truth AUC: 0.804 -> 0.781Mixed
use GradPath instead of straight-line path with GradCF baseline (ablation)IG2 · WM-811K wafer mapSIC AUC-ADD: 0.842 -> 0.898Improved
use GradPath instead of Guided IG path with GradCF baseline (ablation)IG2 · WM-811K wafer mapground truth SUM: 0.472 -> 0.551Improved
use relevant instead of irrelevant references (ablation)IG2 · ImageNet (images with interference objects)SIC AUC-ADD: 0.876 -> 0.901Improved
use relevant instead of irrelevant references (ablation)IG2 · ImageNet (pure background images)SIC AUC-ADD: 0.568 -> 0.537Hurt
use relevant instead of irrelevant references (ablation)IG2 · ImageNet (whole set)ground truth AUC: 0.805 -> 0.798No clear effect
use IG2 instead of Expected IGExpected IG · ImageNetSIC AUC-ADD: 0.557 -> 0.656Improved
use IG2 instead of Expected IGExpected IG · ImageNetSIC AUC-DEL: 0.116 -> 0.115No clear effect
use IG2 instead of Vanilla IGVanilla IG · XAI-BENCH syntheticfaithfulness: 0.349 -> 0.610Improved

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