Research map / Label-efficient and robust learning / Recognition and long-tail
Learning Global Transparent Models consistent with Local Contrastive 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 |
|---|---|---|---|
| train DT on GBFL boolean features instead of original features | decision tree · Sky Survey | C_TB (%): 31.08 -> 48.84 | Helped |
| train DT on GBFL boolean features instead of original features | decision tree · WDBC | C_TB (%): 18.79 -> 50.12 | Helped |
| train DT on GBFL boolean features instead of original features | decision tree · Credit Card | C_TB (%): 3.51 -> 7.03 | Helped |
| train DT on GBFL boolean features instead of original features | decision tree · Magic | C_TB (%): 54.04 -> 56.76 | No clear effect |
| train DT on GBFL boolean features instead of original features | decision tree · Credit Card, Sky Survey, Waveform, WDBC | PN consistency (%): 49.74 -> 74.87 (Credit Card) | Helped |
| train DT on GBFL boolean features instead of original features | decision tree · Sky Survey, Waveform, Magic, Diabetes | PP consistency (%): 75.67 -> 79.44 (Sky Survey) | Mixed |
| train DT on GBFL boolean features instead of original features | decision tree · Sky Survey | test accuracy (%): 99.2 -> 97.68 | Hurt |
| train DT on GBFL boolean features instead of original features | decision tree · Waveform | test accuracy (%): 75.92 -> 75.48 | No clear effect |
| distill black-box predictions into DT instead of training on labels | decision tree · Waveform | C_TB (%): 12.52 -> 5.48 | Hurt |
| augment DT training data with PPs and PNs | decision tree · Sky Survey | C_TB (%): 31.08 -> 3.86 | Hurt |
| augment DT training data with PPs and PNs | decision tree · WDBC | test accuracy (%): 93 -> 95.07 | Helped |
Closest papers by meaning
- Consistent Explanations by Contrastive Learning (CVPR 2022 · similarity 0.89)
- Local vs. Global Interpretability: A Computational Complexity Perspective (ICML 2024 · similarity 0.87)
- Trust Regions for Explanations via Black-Box Probabilistic Certification (ICML 2024 · similarity 0.86)
- From Black-box to Causal-box: Towards Building More Interpretable Models (NeurIPS 2025 · similarity 0.86)
- Keep the Faith: Faithful Explanations in Convolutional Neural Networks for Case-Based Reasoning (AAAI 2024 · similarity 0.85)
- Explainable Neural Networks with Guarantee: A Sparse Estimation Approach (AAAI 2025 · similarity 0.85)
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.