Research map / Label-efficient and robust learning / Semi-supervised learning
From Local Geometry to Global Pseudo-Labeling for Robust Positive–Unlabeled Learning under Covariate Shift
Firas Gabetni, Alexandre Rocchi--Henry, Ziyi LIU, Nacim Belkhir, Gianni Franchi
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 |
|---|---|---|---|
| add classifier-head distillation of kNN pseudo-labels (ablation) | S-PUNA · ImageNet-V2 | AUROC: 92.65 -> 96.38 | Improved |
| add classifier-head distillation of kNN pseudo-labels (ablation) | S-PUNA · ImageNet-V2 | FPR95: 33.70 -> 17.14 | Improved |
| add classifier-head distillation of kNN pseudo-labels (ablation) | S-PUNA · EuroSAT-D (near shift) | AUROC: 84.40 -> 99.79 | Improved |
| add classifier-head distillation of kNN pseudo-labels (ablation) | S-PUNA · ImageNet-1K near shift | FPR95: 19.22 -> 9.69 | Improved |
| add classifier-head distillation of kNN pseudo-labels (ablation) | S-PUNA · TinyImageNet near shift | AUROC: 100.00 -> 99.90 | No clear effect |
| use ViT block 6 features instead of block 11 (ablation) | S-PUNA · ImageNet-1K average | AUROC: 97.03 -> 98.20 | Improved |
| increase shifted proportion of unlabeled pool from 5% to 50% (ablation) | S-PUNA · ImageNet-1K average | AUROC: 91.61 -> 98.20 | Improved |
| increase shifted proportion from 50% to 75% (ablation) | S-PUNA · ImageNet-1K average | AUROC: 98.20 -> 95.53 | Hurt |
| replace Dist-PU with S-PUNA at low shifted proportion | PU baselines · ImageNet-1K average (p=5%) | AUROC: 71.04 -> 91.61 | Improved |
| replace Dist-PU with S-PUNA | PU baselines · ImageNet-1K near shift | FPR95: 35.05 -> 9.69 | Improved |
| replace Dist-PU with S-PUNA | PU baselines · EuroSAT far shift | AUROC: 89.46 -> 100.00 | Improved |
| replace naive kNN propagation with symmetric S-PUNA expansion (no classifier) | kNN baseline · ImageNet-1K near shift | AUROC: 54.40 -> 95.81 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Dist-PU — Dist-PU: Positive-Unlabeled Learning From a Label Distribution Perspective (CVPR 2022) · better on 9 of 9 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Bayesian Adaptation for Covariate Shift (NeurIPS 2021 · similarity 0.85)
- ImageNet-OOD: Deciphering Modern Out-of-Distribution Detection Algorithms (ICLR 2024 · similarity 0.85)
- Generalize or Detect? Towards Robust Semantic Segmentation Under Multiple Distribution Shifts (NeurIPS 2024 · similarity 0.84)
- Computational Efficiency under Covariate Shift in Kernel Ridge Regression (NeurIPS 2025 · similarity 0.83)
- Bridging Multicalibration and Out-of-distribution Generalization Beyond Covariate Shift (NeurIPS 2024 · similarity 0.83)
- High-Dimensional Kernel Methods under Covariate Shift: Data-Dependent Implicit Regularization (ICML 2024 · 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.
