Image Attribution On Vggface2
평가 지표
Deletion AUC score (ArcFace ResNet-101)
Insertion AUC score (ArcFace ResNet-101)
평가 결과
이 벤치마크에서 각 모델의 성능 결과
모델 이름 | Deletion AUC score (ArcFace ResNet-101) | Insertion AUC score (ArcFace ResNet-101) | Paper Title | Repository |
|---|---|---|---|---|
| Kernel SHAP | 0.2034 | 0.6132 | A Unified Approach to Interpreting Model Predictions | - |
| Grad-CAM | 0.3103 | 0.4733 | Grad-CAM: Visual Explanations from Deep Networks via Gradient-based Localization | - |
| HSIC-Attribution | 0.1317 | 0.6694 | Making Sense of Dependence: Efficient Black-box Explanations Using Dependence Measure | - |
| Saliency | 0.1907 | 0.5612 | Deep Inside Convolutional Networks: Visualising Image Classification Models and Saliency Maps | - |
| SMDL-Attribution (ICLR version) | 0.1304 | 0.6705 | Less is More: Fewer Interpretable Region via Submodular Subset Selection | - |
| LIME | 0.2119 | 0.6185 | "Why Should I Trust You?": Explaining the Predictions of Any Classifier | - |
| Integrated Gradients | 0.0749 | 0.5399 | Axiomatic Attribution for Deep Networks | - |
| RISE | 0.1375 | 0.6530 | RISE: Randomized Input Sampling for Explanation of Black-box Models | - |
0 of 8 row(s) selected.