Research map / Label-efficient and robust learning / Recognition and long-tail
Improving Pairwise Ranking for Multi-Label Image Classification
Check 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 hinge-based pairwise ranking with the smooth LSEP loss | VGG16 multi-label classifier · NUS-WIDE | F1 | Helped |
| Use per-class thresholds estimated from image content rather than a global threshold (ablation) | LSEP multi-label classifier · NUS-WIDE | F1: 55.0 -> 52.9 | Hurt |
| Use per-class thresholds estimated from image content rather than a global threshold (ablation) | LSEP multi-label classifier · NUS-WIDE | 0-1 exact match: 29.3 -> 33.5 | Helped |
| Use per-class thresholds estimated from image content rather than fixed top-k selection (ablation) | LSEP multi-label classifier · MS-COCO | 0-1 exact match: 6.19 -> 30.6 | Helped |
| Use per-class thresholds estimated from image content rather than a global threshold (ablation) | LSEP multi-label classifier · MS-COCO | F1: 59.8 -> 62.9 | Helped |
| Use per-class thresholds estimated from image content rather than a global threshold (ablation) | LSEP multi-label classifier · MS-COCO | 0-1 exact match: 23.2 -> 30.6 | Helped |
| Use per-class thresholds estimated from image content rather than a global threshold (ablation) | LSEP multi-label classifier · VOC2007 | F1: 76.4 -> 79.1 | Helped |
| Use per-class thresholds estimated from image content rather than a global threshold (ablation) | LSEP multi-label classifier · VOC2007 | 0-1 exact match: 54.5 -> 64.6 | Helped |
| Add the WARP rank-dependent weighting function to LSEP (ablation) | LSEP loss · Preliminary experiments | performance | No clear effect |
| Train the label prediction and decision modules sequentially rather than jointly (ablation) | LSEP multi-label classifier · Experiments across datasets | performance | Helped |
Closest papers by meaning
- Multi-Label Learning with Pairwise Relevance Ordering (NeurIPS 2021 · similarity 0.90)
- Multi-Label Learning From Single Positive Labels (CVPR 2021 · similarity 0.88)
- Rethinking and Reweighting the Univariate Losses for Multi-Label Ranking: Consistency and Generalization (NeurIPS 2021 · similarity 0.87)
- A Structured Prediction Approach for Label Ranking (NeurIPS 2018 · similarity 0.87)
- MLC-NC: Long-Tailed Multi-Label Image Classification Through the Lens of Neural Collapse (AAAI 2025 · similarity 0.86)
- Double-Layer Hybrid-Label Identification Feature Selection for Multi-View Multi-Label Learning (AAAI 2024 · similarity 0.86)
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.