Research map / Label-efficient and robust learning / Recognition and long-tail
Top-K Pairwise Ranking: Bridging the Gap Among Ranking-Based Measures for Multi-label Classification
Zitai Wang, Qianqian Xu, Zhiyong Yang, Peisong Wen, Yuan He, 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 |
|---|---|---|---|
| replace ranking loss with TKPR loss (alpha2) | L_rank · MS-COCO (ResNet101) | mAP@3: .473 -> .752 | Improved |
| replace ranking loss with TKPR loss (alpha2) | L_rank · MS-COCO (ResNet101) | NDCG@3: .687 -> .813 | Improved |
| replace ranking loss with TKPR loss (alpha2) | L_rank · MS-COCO (ResNet101) | Ranking Loss: .034 -> .024 | Improved |
| replace best pointwise surrogate L_u4 with TKPR (alpha2) | L_u4 · MS-COCO (ResNet101) | mAP@3: .663 -> .752 | Improved |
| replace best pointwise surrogate L_u4 with TKPR (alpha2) | L_u4 · MS-COCO (ResNet101) | P@3: .615 -> .587 | Hurt |
| replace L_u4 with TKPR (alpha2) | L_u4 · MS-COCO (ResNet101) | Ranking Loss: .032 -> .024 | Improved |
| use alpha2 instead of alpha1 in TKPR (ablation) | TKPR · MS-COCO (ResNet101) | mAP@3: .730 -> .752 | Improved |
| use alpha2 instead of alpha3 in TKPR (ablation) | TKPR · MS-COCO (ResNet101) | NDCG@3: .782 -> .813 | Improved |
| use TKPR loss instead of ranking loss for training time (C=10,000, ResNet50) | L_rank · ResNet50, C=10,000 | total time [s]: 0.2306 -> 0.0632 | Improved |
| use TKPR loss instead of torch.ml (C=10,000, ResNet50) | torch.ml · ResNet50, C=10,000 | total time [s]: 0.2907 -> 0.0632 | Improved |
| use TKPR loss instead of pointwise L_u1 for time (C=10,000) | L_u1 · ResNet50, C=10,000 | total time [s]: 0.0510 -> 0.0632 | No clear effect |
| increase classes from C=100 to C=10,000 for TKPR loss | TKPR loss · ResNet50 | total time [s]: 0.0598 -> 0.0632 | No clear effect |
Closest papers by meaning
- Improving Pairwise Ranking for Multi-Label Image Classification (CVPR 2017 · similarity 0.90)
- Multi-Label Learning with Pairwise Relevance Ordering (NeurIPS 2021 · similarity 0.89)
- Multi-label classification: do Hamming loss and subset accuracy really conflict with each other? (NeurIPS 2020 · similarity 0.88)
- Towards a Pairwise Ranking Model with Orderliness and Monotonicity for Label Enhancement (NeurIPS 2025 · similarity 0.87)
- Double-Layer Hybrid-Label Identification Feature Selection for Multi-View Multi-Label Learning (AAAI 2024 · similarity 0.85)
- Decouple then Classify: A Dynamic Multi-view Labeling Strategy with Shared and Specific Information (ICML 2024 · 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.
