Research map / Label-efficient and robust learning / Recognition and long-tail
Measuring the (Un)Faithfulness of Concept-Based Explanations
Shubham Kumar, Narendra Ahuja
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 |
|---|---|---|---|
| randomize concept importances in the explanation (ablation) | SURF surrogate · Caltech-101 ResNet-50 | SURF MAE: 0.00 -> 2.70 | Hurt |
| randomize concept importances in the explanation (ablation) | ICE-Eval · Caltech-101 ResNet-50 | Top-1 agreement: 100 -> 100 | No clear effect |
| randomize concept importances in the explanation (ablation) | C-SHAP-Eval (CEL) · Caltech-101 ResNet-50 | SURF MAE: 1.97 -> 1.97 | No clear effect |
| randomize both CAVs and importances (ablation) | SURF surrogate · Caltech-101 ResNet-50 | Top-1 agreement: 100 -> 1.3 | Hurt |
| randomize both CAVs and importances (ablation) | C-SHAP-Eval (CEL) · Caltech-101 ResNet-50 | Top-1 agreement: 9.02 -> 97.6 | Mixed |
| use HU-MCD instead of MCD for object classification | U-CBEM · ResNet-50 object classification | SURF MAE: 2.60 -> 1.97 | Improved |
| use SAE instead of other U-CBEMs | U-CBEM · ResNet-50 object classification | SURF EMD: 0.384 -> 0.195 | Improved |
| increase the number of concepts | CDISCO, CRAFT, C-SHAP, ICE · object classification | SURF MAE/EMD | No clear effect |
| increase the number of concepts | MCD and HU-MCD · object classification | SURF MAE/EMD | Improved |
Closest papers by meaning
- On the Faithfulness of Vision Transformer Explanations (CVPR 2024 · similarity 0.91)
- On the Faithfulness of Post-Hoc Concept Bottleneck Models (ECCV 2026 · similarity 0.89)
- CoE: Chain-of-Explanation via Automatic Visual Concept Circuit Description and Polysemanticity Quantification (CVPR 2025 · similarity 0.89)
- FaCT: Faithful Concept Traces for Explaining Neural Network Decisions (NeurIPS 2025 · similarity 0.89)
- Probabilistic Conceptual Explainers: Trustworthy Conceptual Explanations for Vision Foundation Models (ICML 2024 · similarity 0.89)
- Measuring Cross-Modal Synergy: A Benchmark for VLM Explainability (NeurIPS 2026 · similarity 0.88)
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.
