Research map / Optimization / Optimizers
When All We Need is a Piece of the Pie: A Generic Framework for Optimizing Two-way Partial AUC
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 |
|---|---|---|---|
| Replace square-loss AUC objective with polynomial surrogate (Poly) for TPAUC optimization | SqAUC · CIFAR-10-LT, CIFAR-100-LT, Tiny-ImageNet-200-LT (9 TPAUC settings) | TPAUC(α,β): e.g. +3.4 on CIFAR-10-LT Subset1 TPAUC(0.3,0.3) | Helped |
| Replace square-loss AUC objective with exponential surrogate (Exp) for TPAUC optimization | SqAUC · CIFAR-10-LT Subset2 and Subset3, CIFAR-100-LT Subset1 | TPAUC(α,β) | Mixed |
| Polynomial surrogate on Tiny-ImageNet-200-LT is not better than SqAUC at TPAUC(0.3,0.3) | SqAUC · Tiny-ImageNet-200-LT Subset3 | TPAUC(0.3,0.3): -0.13 | Hurt |
| Warm-up phase of ordinary AUC training before TPAUC training (Ek epochs) (ablation) | Exp · CIFAR-10-LT Subset2 TPAUC(0.3,0.3) | TPAUC | Helped |
| Warm-up phase of ordinary AUC training before TPAUC training (Ek epochs) (ablation) | Poly · CIFAR-10-LT Subset2 TPAUC(0.3,0.3) | TPAUC | No clear effect |
| Concave weighting function instead of convex weighting | TPAUC surrogate objective · theoretical analysis | upper bound of original risk | Helped |
Closest papers by meaning
- Asymptotically Unbiased Instance-wise Regularized Partial AUC Optimization: Theory and Algorithm (NeurIPS 2022 · similarity 0.91)
- Pairwise AUC Optimization Needs Corrective Power: A Unified View (NeurIPS 2026 · similarity 0.87)
- AUC Optimization from Multiple Unlabeled Datasets (AAAI 2024 · similarity 0.86)
- Interplay of ROC and Precision-Recall AUCs: Theoretical Limits and Practical Implications in Binary Classification (ICML 2024 · similarity 0.85)
- A Closer Look at AUROC and AUPRC under Class Imbalance (NeurIPS 2024 · similarity 0.85)
- Positive-unlabeled AUC Maximization under Covariate Shift (ICML 2025 · 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.