Research map / Learning theory and algorithms / Prediction and decision losses
Statistically Consistent Saliency Estimation
Shunyan Luo, Emre Barut, Fang Jin
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 LEG instead of LIME | saliency estimation on VGG-19 · ImageNet (500 images), Distance-100 masking | average minimal perturbation size S_key: 0.269 -> 0.163 | Improved |
| use LEG instead of KernelSHAP | saliency estimation on VGG-19 · ImageNet (500 images), Distance-100 masking | average minimal perturbation size S_key: 0.251 -> 0.163 | Improved |
| use LEG instead of KernelSHAP | saliency estimation on VGG-19 · ImageNet (500 images), Mean-vector masking | average minimal perturbation size S_key: 0.074 -> 0.118 | Hurt |
| use LEG instead of C-Shapley | saliency estimation on VGG-19 · ImageNet (500 images), Black-out masking | average minimal perturbation size S_key: 0.101 -> 0.062 | Improved |
| add total-variation penalty (LEG-TV vs LEG) (ablation) | LEG · ImageNet (500 images), Black-out masking | average minimal perturbation size S_key: 0.062 -> 0.056 | No clear effect |
| add total-variation penalty (LEG-TV vs LEG) (ablation) | LEG · ImageNet (500 images), Distance-100 masking | average minimal perturbation size S_key: 0.163 -> 0.170 | No clear effect |
| use LEG-TV instead of GradCam | saliency estimation on VGG-19 · ImageNet (500 images), Distance-100 masking | average minimal perturbation size S_key: 0.318 -> 0.170 | Improved |
| use LEG instead of KernelSHAP | saliency estimation on VGG-19 · ImageNet (500 images), Noise masking | average minimal perturbation size S_key: 0.060 -> 0.069 | Hurt |
Closest papers by meaning
- Uncertainty Inspired RGB-D Saliency Detection (TPAMI 2021 · similarity 0.86)
- Saliency Prediction in the Deep Learning Era: Successes and Limitations (TPAMI 2019 · similarity 0.86)
- Sanity Checks for Saliency Maps (NeurIPS 2018 · similarity 0.86)
- Interpretable Explanations of Black Boxes by Meaningful Perturbation (ICCV 2017 · similarity 0.86)
- Rethinking Interpretation: Input-Agnostic Saliency Mapping of Deep Visual Classifiers (AAAI 2023 · similarity 0.86)
- Modeling Saliency Dataset Bias (ICCV 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.
