Research map / Label-efficient and robust learning / Recognition and long-tail
Less Is More: Efficient Black-box Attribution via Minimal Interpretable Subset Selection
Ruoyu Chen, Siyuan Liang, Jingzhi Li, Shiming Liu, Li Liu, Hua Zhang, Xiaochun Cao
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 |
|---|---|---|---|
| use SLICO superpixel division instead of Segment Anything (ablation) | LiMA · ImageNet val, CLIP (ViT-L) | Insertion AUC: 0.7151 -> 0.7441 | Improved |
| use SLICO superpixel division instead of Segment Anything (ablation) | LiMA · ImageNet val, CLIP (ViT-L) | Deletion AUC: 0.1905 -> 0.0926 | Improved |
| use SLICO division instead of SAM (ablation) | LiMA · CUB-200-2011 val | Insertion AUC: 0.7106 -> 0.7748 | Improved |
| use SLICO division instead of SAM (ablation) | LiMA · ImageNet val, LanguageBind | Insertion AUC: 0.6690 -> 0.6058 | Hurt |
| replace previous ICLR version with semantic-division LiMA (SAM) | LiMA · ImageNet val, CLIP (ViT-L) | Insertion AUC: 0.6752 -> 0.7151 | Improved |
| replace RISE with LiMA (SLICO) | LiMA · ImageNet val, ViT-L | Insertion AUC: 0.6177 -> 0.7405 | Improved |
| replace RISE with LiMA (SLICO) | LiMA · ImageNet val, MambaVision (L2) | Insertion AUC: 0.2937 -> 0.3680 | Improved |
| replace HSIC-Attribution with LiMA (SLICO) | LiMA · ImageNet val, LanguageBind | muFidelity: 0.4132 -> 0.4096 | Hurt |
| replace HSIC-Attribution with LiMA (SLICO) | LiMA · ImageNet val, ResNet-101 | muFidelity: 0.5805 -> 0.4203 | Hurt |
| replace HSIC-Attribution with LiMA (SLICO) | LiMA · Celeb-A | Deletion AUC: 0.1151 -> 0.0668 | Improved |
| use bidirectional greedy search instead of standard forward ranking (ablation) | LiMA · ImageNet default N=49 | candidate evaluations: 1,225 -> 805 | Improved |
Closest papers by meaning
- Less is More: Fewer Interpretable Region via Submodular Subset Selection (ICLR 2024 · similarity 0.90)
- GEFA: A General Feature Attribution Framework Using Proxy Gradient Estimation (ICML 2025 · similarity 0.86)
- Provably Better Explanations with Optimized Aggregation of Feature Attributions (ICML 2024 · similarity 0.85)
- Benchmarking the Attribution Quality of Vision Models (NeurIPS 2024 · similarity 0.85)
- Selective Explanations (NeurIPS 2024 · similarity 0.85)
- LibraGrad: Balancing Gradient Flow for Universally Better Vision Transformer Attributions (CVPR 2025 · similarity 0.84)
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.
