Research map / Label-efficient and robust learning / Semi-supervised learning
Weakly Supervised AUC Optimization: A Unified Partial AUC Approach
Zheng Xie, Yu Liu, Hao-Yuan He, Ming Li, Zhi‐Hua Zhou
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 WSAUC instead of AUC-B | noisy label AUC optimization · CIFAR10 (pos 20%, neg 20% noise) | AUC: 88.5 -> 93.6 | Improved |
| use WSAUC instead of AUC-B | noisy label AUC optimization · CIFAR100 (pos 20%, neg 20% noise) | AUC: 78.1 -> 86.4 | Improved |
| use WSAUC instead of AUC-B | noisy label AUC optimization · MNIST (pos 20%, neg 20% noise) | AUC: 99.5 -> 99.6 | No clear effect |
| use WSAUC instead of AUC-B at heavy noise | noisy label AUC optimization · CIFAR100 (pos 40%, neg 40% noise) | AUC: 78.3 -> 76.5 | Hurt |
| use WSAUC instead of DRAUC | noisy label AUC optimization · CIFAR10 (pos 20%, neg 20% noise) | AUC: 38.8 -> 93.6 | Improved |
| use WSAUC instead of PU-AUC | positive-unlabeled AUC optimization · CIFAR10 (pos ratio 20%, label ratio 5%) | AUC: 62.6 -> 76.6 | Improved |
| use WSAUC instead of PU-AUC at 10% labels | positive-unlabeled AUC optimization · CIFAR10 (pos ratio 20%, label ratio 10%) | AUC: 81.7 -> 86.7 | Improved |
| use WSAUC instead of MIDAM (att) | multi-instance AUC optimization · fox | AUC: 73.3 -> 90.4 | Improved |
| use WSAUC instead of Samult | semi-supervised AUC optimization · MNIST (0% noise, 5% label ratio) | AUC: 98.4 -> 98.3 | No clear effect |
| use WSAUC instead of PNU-AUC with scarce noisy labels | semi-supervised AUC optimization · MNIST (30% noise, 5% label ratio) | AUC: 83.2 -> 95.2 | Improved |
Closest papers by meaning
- AUC Optimization from Multiple Unlabeled Datasets (AAAI 2024 · similarity 0.91)
- Semi-Supervised AUC Optimization Without Guessing Labels of Unlabeled Data (AAAI 2018 · similarity 0.91)
- A General Framework for Learning from Weak Supervision (ICML 2024 · similarity 0.86)
- Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency (ICML 2024 · similarity 0.85)
- Weak Supervision Performance Evaluation via Partial Identification (NeurIPS 2024 · similarity 0.85)
- Pairwise AUC Optimization Needs Corrective Power: A Unified View (NeurIPS 2026 · 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.
