Research map / Label-efficient and robust learning / Recognition and long-tail
OpenXAI: Towards a Transparent Evaluation of Model Explanations
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 LIME instead of gradient-based explainers for sign-aware ground-truth faithfulness | feature attribution methods · all datasets (average) | SA and SRA: +61.6% SA, +65.3% SRA | Helped |
| use gradient-based explainers (VanillaGrad, SmoothGrad, IntegratedGrad) instead of random | feature attribution methods · all datasets | PRA, FA, RA, RC: RC 0.005 -> 1.0 | Helped |
| use SmoothGrad instead of other explainers for predictive faithfulness | feature attribution methods · multiple datasets (average) | PGU: +43.03% | Helped |
| use Gradient x Input instead of other explainers | feature attribution methods · German credit, ANN | predictive faithfulness: +6.74% | Helped |
| use Gradient x Input instead of other explainers for stability | feature attribution methods · Synthetic, LR | RIS and RRS: +93.5% RIS, +59.2% RRS | Helped |
| use Gradient x Input on real-world instead of synthetic data | Gradient x Input · real-world datasets | stability | Hurt |
| use SmoothGrad instead of other explainers for representation stability | feature attribution methods · real-world datasets (average) | RRS: +63.2% | Helped |
| choose among explainers for relative input stability | feature attribution methods · real-world datasets | RIS | No clear effect |
| use Gradient x Input instead of other explainers for fairness | feature attribution methods · German Credit and Adult, LR | fairness of PGU: +8.9% | Helped |
Closest papers by meaning
- Representational Difference Explanations (NeurIPS 2025 · similarity 0.86)
- Evaluate with the Inverse: Efficient Approximation of Latent Explanation Quality Distribution (AAAI 2025 · similarity 0.85)
- Navigating the Maze of Explainable AI: A Systematic Approach to Evaluating Methods and Metrics (NeurIPS 2024 · similarity 0.85)
- OpenCUA: Open Foundations for Computer-Use Agents (NeurIPS 2025 · similarity 0.85)
- BEE: Metric-Adapted Explanations via Baseline Exploration-Exploitation (AAAI 2025 · similarity 0.84)
- Faithful Model Explanations through Energy-Constrained Conformal Counterfactuals (AAAI 2024 · similarity 0.84)
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.