Research map / Label-efficient and robust learning / Recognition and long-tail
Learning from Positive and Unlabeled Data with Arbitrary Positive Shift
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 |
|---|---|---|---|
| Use PU2 wUU instead of PUc under fully disjoint positive supports | aPU learning · MNIST (N: 0,2; Ptest: 5,7; Ptrain: 1,3) | misclassification rate (%): 17.1 -> 3.1 | Helped |
| Use PU2 wUU instead of PUc under fully disjoint positive supports | aPU learning · 20 Newsgroups (N: misc,rec; Ptest: soc,talk; Ptrain: alt,comp) | misclassification rate (%): 18.5 -> 5.6 | Helped |
| Use PU2 wUU instead of PUc under fully disjoint positive supports | aPU learning · CIFAR10 (N: Deer,Horse; Ptest: Plane,Auto; Ptrain: Cat,Dog) | misclassification rate (%): 33.1 -> 11.2 | Helped |
| Use PU2 wUU instead of PUc under adversarial spam drift | aPU learning · TREC05 to TREC07 spam (train prior 0.4) | misclassification rate (%): 35.2 -> 25.1 | Helped |
| Reduce the training positive prior | PURR, PU2aPNU, and PU2 wUU · Shifted MNIST, 20 Newsgroups, and CIFAR10 tasks | misclassification rate (%): null | Helped |
| Reduce the training positive prior | PUc · Shifted MNIST, 20 Newsgroups, and CIFAR10 tasks | misclassification rate (%): null | Hurt |
| Use PU2aPNU rather than PU2 wUU for partially disjoint positive supports | Two-step aPU learning · Partially disjoint positive-support tasks | misclassification rate (%): null | Helped |
| Use PU2aPNU rather than PU2 wUU for fully disjoint positive supports | Two-step aPU learning · Fully disjoint positive-support tasks | misclassification rate (%): null | Hurt |
| Use the two-step learners rather than PURR | aPU learning · Reported shifted tasks | misclassification rate (%): null | Mixed |
Closest papers by meaning
- Positive-Unlabeled Learning with Non-Negative Risk Estimator (NeurIPS 2017 · similarity 0.94)
- Balancing Positive and Negative Classification Error Rates in Positive-Unlabeled Learning (NeurIPS 2025 · similarity 0.93)
- Learning from Positive and Unlabeled Data with a Selection Bias (ICLR 2019 · similarity 0.92)
- Classification from Positive, Unlabeled and Biased Negative Data (ICML 2019 · similarity 0.92)
- Learning from positive and unlabeled examples -Finite size sample bounds (NeurIPS 2025 · similarity 0.91)
- Positive and Unlabeled Learning with Controlled Probability Boundary Fence (ICML 2024 · similarity 0.90)
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.