Image Attribution On Celeba
评估指标
Deletion AUC score (ArcFace ResNet-101)
Insertion AUC score (ArcFace ResNet-101)
评测结果
各个模型在此基准测试上的表现结果
| Paper Title | Repository | |||
|---|---|---|---|---|
| Grad-CAM | 0.2865 | 0.3721 | Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization | |
| LIME | 0.1484 | 0.5246 | "Why Should I Trust You?": Explaining the Predictions of Any Classifier | |
| Saliency | 0.1453 | 0.4632 | Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps | |
| RISE | 0.1444 | 0.5703 | RISE: Randomized Input Sampling for Explanation of Black-box Models | |
| Kernel SHAP | 0.1409 | 0.5246 | A Unified Approach to Interpreting Model Predictions | |
| HSIC-Attribution | 0.1151 | 0.5692 | Making Sense of Dependence: Efficient Black-box Explanations Using Dependence Measure | |
| SMDL-Attribution (ICLR version) | 0.1054 | 0.5752 | Less is More: Fewer Interpretable Region via Submodular Subset Selection | |
| Integrated Gradients | 0.0680 | 0.3578 | Axiomatic Attribution for Deep Networks |
0 of 8 row(s) selected.