Research map / Learning theory and algorithms / Prediction and decision losses
A Bayesian Framework for Learning Rule Sets for Interpretable Classification
Tong Wang, Cynthia Rudin, Finale Doshi-Velez, Yimin Liu, Erica Klampfl, Perry MacNeille
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 |
|---|---|---|---|
| use BRS instead of C4.5 | BRS-BetaBinomial · UCI tic-tac-toe | accuracy: .92 -> 1.00 | Improved |
| use BRS instead of Lasso | BRS-BetaBinomial · UCI tic-tac-toe | accuracy: .71 -> 1.00 | Improved |
| use BRS instead of C4.5 | BRS-BetaBinomial · UCI connect-4 | accuracy: .83 -> .76 | Hurt |
| use BRS instead of random forest | BRS-BetaBinomial · UCI chess | accuracy: .91 -> .89 | Hurt |
| use BRS instead of CART | BRS-BetaBinomial · In-vehicle coupons: coffee house | AUC: 0.615 -> 0.682 | Improved |
| use BRS instead of TopK rule ranking | BRS-BetaBinomial · In-vehicle coupons: bar | AUC: 0.562 -> 0.773 | Improved |
| use Lasso instead of BRS | Lasso · In-vehicle coupons: bar | AUC: 0.773 -> 0.795 | Improved |
| increase prior parameter rho toward 1 (ablation) | BRS-BetaBinomial · UCI tic-tac-toe, diabetes, mixed set | accuracy | Improved |
| vary likelihood scale s among 100, 1000, 10000 (ablation) | BRS-BetaBinomial · UCI parameter study | accuracy | No clear effect |
| raise minimum support using the Theorem 1 bound (ablation) | BRS-BetaBinomial · tic-tac-toe rule space | fraction of rules remaining: 100% -> 9.0% | Improved |
| strengthen prior (beta_l = 10|A_l|) for small models (ablation) | BRS-BetaBinomial · tic-tac-toe rule space | minimum support: 9 -> 13 | Improved |
| increase number of observations N or rules M (ablation) | BRS-BetaBinomial · simulation study 1 | training error convergence | No clear effect |
Closest papers by meaning
- Multi-value Rule Sets for Interpretable Classification with Feature-Efficient Representations (NeurIPS 2018 · similarity 0.86)
- Scalable Bayesian Rule Lists (ICML 2017 · similarity 0.85)
- Learning Interpretable Rules for Scalable Data Representation and Classification (TPAMI 2023 · similarity 0.84)
- Would You Like Your Data to Be Trained? A User Controllable Recommendation Framework (AAAI 2024 · similarity 0.83)
- Bayesian Inference for Correlated Human Experts and Classifiers (ICML 2025 · similarity 0.83)
- Influential Exemplar Replay for Incremental Learning in Recommender Systems (AAAI 2024 · similarity 0.82)
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.
