Research map / Learning theory and algorithms / Prediction and decision losses
Closing the Approximation Gap of Partial AUC Optimization: A Tale of Two Formulations
Yangbangyan Jiang, Qianqian Xu, Huiyang Shao, Zhiyong Yang, Shilong Bao, Xiaochun Cao, Qingming Huang
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 unbiased reformulation (UPAUCI) instead of vanishing-gap approximation (PAUCI) | PAUCI · iNaturalist2021 OPAUC FPR<=0.3 | OPAUC: 0.9348 -> 0.9587 | Improved |
| use unbiased reformulation (UPAUCI) instead of PAUCI | PAUCI · iNaturalist2021 TPAUC (0.3,0.3) | TPAUC: 0.8219 -> 0.8458 | Improved |
| use unbiased reformulation (UPAUCI) instead of PAUCI | PAUCI · Tiny-ImageNet-200-LT-1 OPAUC FPR<=0.3 | OPAUC: 0.8267 -> 0.8559 | Improved |
| use unbiased reformulation (UPAUCI) instead of PAUCI | PAUCI · Tiny-ImageNet-200-LT-1 TPAUC (0.3,0.3) | TPAUC: 0.4606 -> 0.4927 | Improved |
| use unbiased reformulation (UPAUCI) instead of PAUCI | PAUCI · CIFAR-10-LT-3 OPAUC FPR<=0.3 | OPAUC: 0.7746 -> 0.8061 | Improved |
| use unbiased reformulation (UPAUCI) instead of PAUCI | PAUCI · CIFAR-100-LT-3 TPAUC (0.3,0.3) | TPAUC: 0.5027 -> 0.5465 | Improved |
| use unbiased reformulation (UPAUCI) instead of PAUCI | PAUCI · CIFAR-10-LT-2 TPAUC (0.3,0.3) | TPAUC: 0.9021 -> 0.9015 | No clear effect |
| use instance-wise PAUCI instead of AUC optimization (AUC-M) | AUC-M · Tiny-ImageNet-200-LT-1 OPAUC FPR<=0.3 | OPAUC: 0.8102 -> 0.8267 | Improved |
| use unbiased UPAUCI instead of biased approximations AUC-poly | AUC-poly · Tiny-ImageNet-200-LT-1 OPAUC FPR<=0.3 | OPAUC: 0.8185 -> 0.8559 | Improved |
| use UPAUCI instead of SOPA on CIFAR-100-LT-1 OPAUC FPR<=0.5 | SOPA · CIFAR-100-LT-1 OPAUC FPR<=0.5 | OPAUC: 0.9275 -> 0.9221 | Hurt |
| initialize c larger (1 instead of 0.2) (ablation) | UPAUCI · CIFAR-10-LT-1 OPAUC FPR<=0.5 | OPAUC: gap approaching 0.1 | Improved |
| add small strongly-concave squared term with omega (vs omega=0) (ablation) | UPAUCI · CIFAR-10-LT-1 | OPAUC/TPAUC | Improved |
Closest papers by meaning
- Asymptotically Unbiased Instance-wise Regularized Partial AUC Optimization: Theory and Algorithm (NeurIPS 2022 · similarity 0.93)
- A Closer Look at AUROC and AUPRC under Class Imbalance (NeurIPS 2024 · similarity 0.86)
- Interplay of ROC and Precision-Recall AUCs: Theoretical Limits and Practical Implications in Binary Classification (ICML 2024 · similarity 0.86)
- Positive-unlabeled AUC Maximization under Covariate Shift (ICML 2025 · similarity 0.86)
- AUC Maximization under Positive Distribution Shift (NeurIPS 2024 · similarity 0.85)
- A Novel Characterization of the Population Area Under the Risk Coverage Curve (AURC) and Rates of Finite Sample Estimators (ICML 2025 · similarity 0.85)
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.
