Research map / Label-efficient and robust learning / Recognition and long-tail
Shap-CAM: Visual Explanations for Convolutional Neural Networks Based on Shapley Value
Quan Zheng, Ziwei Wang, Jie Zhou, Jiwen Lu
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 |
|---|---|---|---|
| replace Score-CAM with Shap-CAM | VGG16 CAM · ImageNet val | Average Drop (%): 31.5 -> 28.0 | Improved |
| replace Score-CAM with Shap-CAM | VGG16 CAM · ImageNet val | Average Increase in Confidence (%): 30.6 -> 31.8 | Improved |
| replace Score-CAM with Shap-CAM | VGG16 CAM · PASCAL VOC 2007 val | Average Drop (%): 15.6 -> 13.2 | Improved |
| replace Score-CAM with Shap-CAM | VGG16 CAM · PASCAL VOC 2007 val | Average Increase in Confidence (%): 28.9 -> 32.7 | Improved |
| replace Grad-CAM with Shap-CAM | VGG16 CAM · ImageNet val | Average Drop (%): 47.8 -> 28.0 | Improved |
| replace Score-CAM with Shap-CAM | VGG-16 · energy-based pointing game | proportion (%): 40.10 -> 40.45 | Improved |
| replace Score-CAM with Shap-CAM | ResNet18 · energy-based pointing game | proportion (%): 40.76 -> 41.28 | Improved |
| add Shap-CAM explanation loss to student training (without KD loss) | WRN-16-2 student · CIFAR-10 | test error rate (%): 6.78 -> 6.69 | Improved |
| add Shap-CAM explanation loss to student training (with KD loss) | WRN-16-2 student · CIFAR-10 | test error rate (%): 5.68 -> 5.37 | Improved |
| add Grad-CAM explanation loss to student training (without KD loss) | WRN-16-2 student · CIFAR-10 | test error rate (%): 6.78 -> 6.86 | Hurt |
| add Grad-CAM explanation loss to student training (with KD loss) | WRN-16-2 student · CIFAR-10 | test error rate (%): 5.68 -> 5.80 | Hurt |
Closest papers by meaning
- Towards Better Explanations of Class Activation Mapping (ICCV 2021 · similarity 0.93)
- DD-CAM: Minimal Sufficient Explanations for Vision Models Using Delta Debugging (NeurIPS 2026 · similarity 0.88)
- Empowering CAM-Based Methods with Capability to Generate Fine-Grained and High-Faithfulness Explanations (AAAI 2024 · similarity 0.87)
- Finer-CAM: Spotting the Difference Reveals Finer Details for Visual Explanation (CVPR 2025 · similarity 0.87)
- CAPE: CAM as a Probabilistic Ensemble for Enhanced DNN Interpretation (CVPR 2024 · similarity 0.86)
- DiffCAM: Data-Driven Saliency Maps by Capturing Feature Differences (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.
