Research map / Label-efficient and robust learning / Recognition and long-tail
Neyman-Pearson Classification under High-Dimensional Settings
Anqi Zhao, Yang Feng, Lie Wang, Xin Tong
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 PSN2 instead of nb, pen-log, svm to control type I error | Neyman-Pearson classifier · Neuroblastoma | type I error: .308 -> .038 | Improved |
| use PSN2 instead of pen-log | Neyman-Pearson classifier · p53 mutants | type I error: .162 -> .019 | Improved |
| use PSN2 instead of pen-log | Neyman-Pearson classifier · p53 mutants | type II error: .010 -> .461 | Hurt |
| use NSN2-R0 instead of svm | Neyman-Pearson classifier · Email spam + noise | type I error: .203 -> .019 | Improved |
| switch NSN2 priority from type I to type II control | NSN2 · Email spam + noise | type II error: .439 -> .020 | Improved |
| add screening to the Neyman-Pearson Naive Bayes classifier | NSN2 / PSN2 · simulated d=1000 | type II error | Improved |
| add screening to the classifier at large sample sizes | PSN2 · simulated d=100 | type II error | No clear effect |
| add screening when all features are signals | PN2 · simulated d=10 | type II error | Hurt |
| use D-statistic instead of t-statistic screening on non-normal data | NSN2 · Example 2 (d=1000) | missed signals: 9.50 -> 1.26 | Improved |
| use D-statistic instead of t-statistic screening on normal data | screening · Example 1 (d=1000) | missed signals: 0.48 -> 1.14 | Hurt |
| use nonparametric NSN2 instead of parametric PSN2 | Neyman-Pearson Naive Bayes · Example 2 non-normal | error | Improved |
| use parametric instead of nonparametric estimation | Neyman-Pearson Naive Bayes · Example 1 Gaussian | type II error | Improved |
Closest papers by meaning
- Non-splitting Neyman-Pearson Classifiers (JMLR 2024 · similarity 0.93)
- Neyman-Pearson classification: parametrics and sample size requirement (JMLR 2020 · similarity 0.93)
- Optimal Binary Classification Beyond Accuracy (NeurIPS 2022 · similarity 0.85)
- Principled Algorithms for Optimizing Generalized Metrics in Binary Classification (ICML 2025 · similarity 0.82)
- Concentration and excess risk bounds for imbalanced classification with synthetic oversampling (NeurIPS 2025 · similarity 0.82)
- The VOROS: Lifting ROC Curves to 3D to Summarize Unbalanced Classifier Performance (AAAI 2025 · 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.
