Research map / Architectures and efficiency / RNNs and normalization
Learn to Rank: Visual Attribution by Learning Importance Ranking
David Schinagl, Christian Fruhwirth-Reisinger, Alexander Prutsch, Samuel Schulter, Horst Possegger
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add 3 test-time refinement steps (ablation) | AHA · ImageNet val, ViT-B/16 classifier, predicted class | Deletion AUC: 0.1381 -> 0.1215 | Improved |
| add 3 test-time refinement steps (ablation) | AHA · ImageNet val, ViT-B/16 classifier, predicted class | Insertion AUC: 0.6112 -> 0.6485 | Improved |
| increase training perturbation steps S from 4 to 16 (ablation) | AHA · ImageNet val, ViT-B/16 | Insertion AUC | Improved |
| increase training perturbation steps beyond 16 (ablation) | AHA · ImageNet val, ViT-B/16 | Insertion AUC | No clear effect |
| sample grid size G from U(7,28) instead of fixed G=7 (ablation) | AHA · ViT-B/16 hyperparameter ablation | RISE Ins.-Del.: 0.671 -> 0.715 | Improved |
| vary sorting temperature tau (ablation) | AHA · ViT-B/16 hyperparameter ablation | RISE Ins.-Del.: 0.703 to 0.715 | No clear effect |
| optimize deletion/insertion directly with learned amortized explainer instead of per-sample optimization (MDA) | AHA · MDA protocol (5,000 images), ViT-B/16 | Insertion - Deletion: 0.624 -> 0.716 | Improved |
| use learned explainer instead of TIS on insertion at T=0 | AHA · ImageNet val, ViT-B/16 classifier, predicted class | Insertion AUC: 0.6381 -> 0.6112 | Hurt |
| apply explainer trained on ViT-B/32 to ViT-B/16 with T=3 refinement (ablation) | AHA · Cross-architecture transfer B/32 to B/16 (RISE Ins.-Del.) | RISE Ins.-Del.: 0.650 -> 0.739 | Improved |
| apply AHA to ConvNeXt without architecture changes | AHA · ImageNet val, ConvNeXt Small classifier, predicted class | Deletion AUC: 0.2748 -> 0.1741 | Improved |
| AHA T=0 versus MDA on segmentation localization | AHA · ImageNet segmentation, ViT-B/32 | IoU: 0.702 -> 0.667 | Hurt |
Closest papers by meaning
- Benchmarking the Attribution Quality of Vision Models (NeurIPS 2024 · similarity 0.89)
- Interpreting Object-level Foundation Models via Visual Precision Search (CVPR 2025 · similarity 0.87)
- Probabilistic Conceptual Explainers: Trustworthy Conceptual Explanations for Vision Foundation Models (ICML 2024 · similarity 0.87)
- LibraGrad: Balancing Gradient Flow for Universally Better Vision Transformer Attributions (CVPR 2025 · similarity 0.86)
- Pixel-level Certified Explanations via Randomized Smoothing (ICML 2025 · similarity 0.86)
- Comprehensive Information Bottleneck for Unveiling Universal Attribution to Interpret Vision Transformers (CVPR 2025 · similarity 0.86)
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.
