Research map / Label-efficient and robust learning / Recognition and long-tail
Distributionally Robust Loss for Long-Tailed Multi-Label Image Classification
Dekun Lin, Zhe Cui, Rui Chen, Tailai Peng, xinran xie, Xiaolin Qin
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 |
|---|---|---|---|
| compute LSEP class-wise (C-LSEP) instead of per sample (ablation) | LSEP · VOC-LT | total mAP: 72.99 -> 74.93 | Improved |
| compute LSEP class-wise (C-LSEP) instead of per sample (ablation) | LSEP · COCO-LT | total mAP: 47.05 -> 50.35 | Improved |
| add negative gradient constraint (NGC) to C-LSEP (ablation) | C-LSEP · VOC-LT | total mAP: 74.93 -> 78.01 | Improved |
| add negative gradient constraint (NGC) to C-LSEP (ablation) | C-LSEP · COCO-LT | total mAP: 50.35 -> 54.10 | Improved |
| add positive gradient constraint (PGC) to C-LSEP (ablation) | C-LSEP · VOC-LT | total mAP: 74.93 -> 74.74 | Hurt |
| add PGC on top of NGC (GC) to C-LSEP (ablation) | C-LSEP + NGC · VOC-LT | total mAP: 78.01 -> 76.90 | Hurt |
| add PGC on top of NGC (GC) to C-LSEP (ablation) | C-LSEP + NGC · COCO-LT | total mAP: 54.10 -> 53.58 | Hurt |
| compute ZLPR class-wise (C-ZLPR) (ablation) | ZLPR · COCO-LT | total mAP: 49.90 -> 51.76 | Improved |
| combine DR loss with DB Focal (ablation) | DR Loss · VOC-LT | total mAP: 78.01 -> 78.75 | Improved |
| combine DR loss with DB Focal (ablation) | DR Loss · COCO-LT | total mAP: 54.10 -> 54.55 | Improved |
| combine baseline C-LSEP with DB Focal (ablation) | DB Focal · VOC-LT | total mAP: 78.17 -> 76.21 | Hurt |
| combine BCE with DR loss (ablation) | DR Loss · COCO-LT | total mAP: 54.10 -> 52.84 | Hurt |
Closest papers by meaning
- Distribution-Balanced Loss for Multi-Label Classification in Long-Tailed Datasets (ECCV 2020 · similarity 0.93)
- Two-Way Multi-Label Loss (CVPR 2023 · similarity 0.92)
- Multi-Label Learning with Stronger Consistency Guarantees (NeurIPS 2024 · similarity 0.89)
- MLC-NC: Long-Tailed Multi-Label Image Classification Through the Lens of Neural Collapse (AAAI 2025 · similarity 0.89)
- Semi-supervised Multi-label Learning with Balanced Binary Angular Margin Loss (NeurIPS 2024 · similarity 0.88)
- Harnessing Hierarchical Label Distribution Variations in Test Agnostic Long-tail Recognition (ICML 2024 · similarity 0.87)
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.
