AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

ConEx: Human-Interpretable Saliency Maps via Concept-Aware Attribution

Yehonatan Elisha, Oren Barkan, Ziv Weiss Haddad, Noam Koenigstein

ICML 2026 · Vision-language and multimodal › Human and embodied modeling · approach: Language-grounded visual concept learning

Read the paper

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
build CAVs from concept segments instead of full images (ablation)ConEx · ImageNet (RN, CN, DN)VCM: 0.59 -> 0.68Improved
build CAVs from concept segments instead of full images (ablation)ConEx · ImageNet (RN, CN, DN)CCM: 24.68 -> 27.83Improved
use ConEx centroid-difference CAV construction instead of MCD's CAV construction (ablation)ConEx · ImageNet RNINS: 58.16 -> 59.21Improved
use ConEx CAV construction instead of ACE's (ablation)ConEx · ImageNet RNDEL: 9.58 -> 8.34Improved
use LRP multiplicative fusion instead of GradCAM fusion (ablation)ConEx · ImageNet RNINS: 58.68 -> 59.21Improved
use LRP multiplicative fusion instead of SHAP fusion (ablation)ConEx · ImageNet RNINS: 56.99 -> 59.21Improved
use LRP multiplicative fusion instead of IIA fusion (ablation)ConEx · ImageNet RNDEL: 8.65 -> 8.34Improved
wrap GradCAM with the ConEx concept frameworkGradCAM · ImageNet RNINS: 54.92 -> 58.68Improved
wrap SHAP with the ConEx concept frameworkSHAP · ImageNet RNINS: 47.54 -> 56.99Improved
wrap IIA with the ConEx concept frameworkIIA · ImageNet RNDEL: 9.48 -> 8.65Improved
use the last layer instead of layer 12 -> middle layer L=6 for ViT-B segment embeddings (ablation)ConEx · ImageNet ViT-BCINS: 32.40 -> 35.38Improved
use last layer L instead of L-2 for RN segment embeddings (ablation)ConEx · ImageNet RNCCM: 27.56 -> 28.19Improved

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