Research map / Learning theory and algorithms / Prediction and decision losses
Positive-Unlabeled Learning using Random Forests via Recursive Greedy Risk Minimization
Jonathan Wilton, Abigail Koay, Ryan Ko, Miao Xu, Nan Ye
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 nnPU instead of uPU risk estimator (quadratic loss) (ablation) | PU ET · MNIST | accuracy (%): 50.74 -> 93.60 | Improved |
| use nnPU instead of uPU risk estimator (quadratic loss) (ablation) | PU ET · 20News | accuracy (%): 43.64 -> 83.34 | Improved |
| use logistic instead of quadratic loss with uPU risk (ablation) | PU ET · 20News | accuracy (%): 43.64 -> 73.44 | Improved |
| use logistic instead of quadratic loss with nnPU risk (ablation) | PU ET · MNIST | accuracy (%): 93.60 -> 94.01 | No clear effect |
| use logistic instead of quadratic loss with nnPU risk (ablation) | PU ET · Covtype | accuracy (%): 76.51 -> 75.63 | Mixed |
| PU ET versus treating unlabeled as negative (NaivePUET) | PU ET · MNIST | accuracy (%): 50.74 -> 93.60 | Improved |
| PU ET versus PUBaggingET | PU ET · Mushroom | accuracy (%): 62.76 -> 99.70 | Improved |
| PU ET versus nnPU neural net | PU ET · 20News | accuracy (%): 74.11 -> 83.34 | Improved |
| PU ET versus nnPU neural net | PU ET · Epsilon | accuracy (%): 73.80 -> 57.39 | Hurt |
| PU ET versus Self-PU | PU ET · CIFAR-10 | accuracy (%): 88.22 -> 79.74 | Hurt |
| PU ET versus fully supervised ET | PU ET · Covtype | accuracy (%): 95.90 -> 76.51 | Hurt |
Closest papers by meaning
- Positive-Unlabeled Learning with Non-Negative Risk Estimator (NeurIPS 2017 · similarity 0.89)
- Risk Bounds for Positive-Unlabeled Learning Under the Selected At Random Assumption (JMLR 2023 · similarity 0.88)
- Learning from positive and unlabeled examples -Finite size sample bounds (NeurIPS 2025 · similarity 0.87)
- Learning Decision Trees and Forests with Algorithmic Recourse (ICML 2024 · similarity 0.86)
- Robust Loss Functions for Training Decision Trees with Noisy Labels (AAAI 2024 · similarity 0.86)
- Learning Small Decision Trees with Few Outliers: A Parameterized Perspective (AAAI 2024 · similarity 0.86)
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.
