Research map / Learning theory and algorithms / Prediction and decision losses
Consistency Analysis for Binary Classification Revisited
Krzysztof Dembczyński, Wojciech Kotłowski, Oluwasanmi Koyejo, Nagarajan Natarajan
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 |
|---|---|---|---|
| Recalibrate logistic CPE versus plain logistic in ETU (ablation) | ETU · SPAMBASE | F1: 0.8448 -> 0.8905 | Improved |
| Recalibrate logistic CPE versus plain logistic in ETU (ablation) | ETU · BREAST CANCER | F1: 0.9660 -> 0.9799 | Improved |
| Recalibrate logistic CPE versus plain logistic in ETU (ablation) | ETU · WEB PAGE | F1: 0.6336 -> 0.6756 | Improved |
| Recalibrate logistic CPE versus plain logistic in ETU (ablation) | ETU · SCENE | F1: 0.6916 -> 0.6809 | No clear effect |
| Isotron CPE versus logistic CPE in ETU (ablation) | ETU · SCENE | F1: 0.6916 -> 0.6222 | Hurt |
| Isotron CPE versus logistic CPE in ETU (ablation) | ETU · SPAMBASE | F1: 0.8448 -> 0.7825 | Hurt |
| Isotron CPE versus logistic CPE in ETU (ablation) | ETU · WEB PAGE | F1: 0.6336 -> 0.7362 | Improved |
| Second-order versus first-order approximation of ETU (ablation) | ETU approximation · Synthetic F1 and Jaccard | approximation error | Improved |
| First-order approximation instead of exact ETU (ablation) | ETU approximation · SPAMBASE | F1: 0.8448 -> 0.8457 | No clear effect |
| Second-order approximation instead of exact ETU (ablation) | ETU approximation · LETTERS | Jaccard: 0.4273 -> 0.4273 | No clear effect |
| PU algorithm versus recalibrated logistic ETU | PU · Most datasets | F1: 0.8905 -> 0.8839 | No clear effect |
| PU versus recalibrated logistic ETU | PU · SCENE | F1: 0.6809 -> 0.4397 | Hurt |
Closest papers by meaning
- Principled Algorithms for Optimizing Generalized Metrics in Binary Classification (ICML 2025 · similarity 0.88)
- Consistent algorithms for multi-label classification with macro-at-$k$ metrics (ICLR 2024 · similarity 0.84)
- Don’t Label Twice: Quantity Beats Quality when Comparing Binary Classifiers on a Budget (ICML 2024 · similarity 0.84)
- Simple Weak Coresets for Non-decomposable Classification Measures (AAAI 2024 · similarity 0.84)
- Trading off Consistency and Dimensionality of Convex Surrogates for Multiclass Classification (NeurIPS 2024 · similarity 0.84)
- Accuracy vs. Accuracy: Computational Tradeoffs Between Classification Rates and Utility (NeurIPS 2026 · similarity 0.83)
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.
