Research map / Learning theory and algorithms / Prediction and decision losses
Constraint-Driven Explanations for Black-Box ML Models
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 |
|---|---|---|---|
| Add a user constraint limiting explanations to clients not previously contacted, alongside integrity constraints (ablation) | CLIME · Bank dataset | fidelity: 0.89 -> 0.98 | Helped |
| Apply integrity constraints rather than sampling without constraints (ablation) | CLIME · Bank dataset | fidelity: 0.90 -> 0.89 | Hurt |
| Add a constraint that capital gain and capital loss are UNKNOWN (ablation) | CLIME · Adult dataset, male individuals without a college degree | fidelity: 0.68 -> 0.98 | Helped |
| Use the CLIME fidelity estimator instead of ApproxMC | Fidelity estimation on Anchor benchmarks · 150 Anchor benchmarks | runtime: 7.5× faster | Helped |
| Constrain samples to low Hamming distance from nearby instances | CLIME · Recidivism adversarial classifier | frequency of race appearing as top explanation: 60% | Helped |
| Vary Hamming-distance constraints on a non-adversarial classifier (ablation) | CLIME · Recidivism non-adversarial classifier | relative frequency of a feature appearing as top explanation | No clear effect |
Closest papers by meaning
- Learning with Explanation Constraints (NeurIPS 2023 · similarity 0.91)
- Faithful Model Explanations through Energy-Constrained Conformal Counterfactuals (AAAI 2024 · similarity 0.87)
- Trust Regions for Explanations via Black-Box Probabilistic Certification (ICML 2024 · similarity 0.87)
- Efficient and Accurate Explanation Estimation with Distribution Compression (ICLR 2025 · similarity 0.86)
- Towards Modeling Uncertainties of Self-Explaining Neural Networks via Conformal Prediction (AAAI 2024 · similarity 0.86)
- Enhancing Performance of Explainable AI Models with Constrained Concept Refinement (ICML 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.