Research map / Probabilistic and scientific modeling / Variational and generative sampling
Positive Distribution Pollution: Rethinking Positive Unlabeled Learning from a Unified Perspective
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 rebalanced risk estimation (ablation) | CoVPU · MNIST | F1: 0.971 -> 0.985 | Helped |
| add rebalanced risk estimation (ablation) | CoVPU · Cifar10 | F1: 0.520 -> 0.527 | Helped |
| add rebalanced risk estimation (ablation) | CoVPU · BankLoan | F1: 0.416 -> 0.447 | Helped |
| add VPU-Calibrator (ablation) | CoVPU · MNIST | F1: 0.950 -> 0.985 | Helped |
| add VPU-Calibrator (ablation) | CoVPU · Cifar10 | F1: 0.520 -> 0.527 | Helped |
| add VPU-Calibrator (ablation) | CoVPU · BankLoan | F1: 0.392 -> 0.447 | Helped |
| use prior-free variational loss instead of prior-based risk estimation (ablation) | CoVPU · MNIST | F1: 0.966 -> 0.985 | Helped |
| use prior-free variational loss instead of prior-based risk estimation (ablation) | CoVPU · Cifar10 | F1: 0.522 -> 0.527 | Helped |
| use prior-free variational loss instead of prior-based risk estimation (ablation) | CoVPU · BankLoan | F1: 0.446 -> 0.447 | No clear effect |
| vary the supplied class prior away from the true prior | prior-based PU methods · Cifar10 and MYbank | F1 | Hurt |
Closest papers by meaning
- Learning from Positive and Unlabeled Data with Arbitrary Positive Shift (NeurIPS 2020 · similarity 0.89)
- Positive-Unlabeled Learning with Non-Negative Risk Estimator (NeurIPS 2017 · similarity 0.88)
- Positive-Unlabeled Learning with Extreme Scarcity of Labeled Positives (ICML 2026 · similarity 0.88)
- Balancing Positive and Negative Classification Error Rates in Positive-Unlabeled Learning (NeurIPS 2025 · similarity 0.87)
- Learning from positive and unlabeled examples -Finite size sample bounds (NeurIPS 2025 · similarity 0.85)
- Sample-Optimal Agnostic Boosting with Unlabeled Data (ICML 2025 · similarity 0.84)
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.