Research map / Label-efficient and robust learning / Recognition and long-tail
Learning in Imperfect Environment: Multi-Label Classification with Long-Tailed Distribution and Partial Labels
Wenqiao Zhang, Changshuo Liu, Lingze Zeng, Bengchin Ooi, Siliang Tang, Yueting Zhuang
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 |
|---|---|---|---|
| add label correction (RLC) (ablation) | COMIC · PLT-COCO | Total mAP: 54.70 -> 55.08 | Improved |
| add multi-focal modifier loss (ablation) | COMIC · PLT-COCO | Total mAP: 54.60 -> 55.08 | Improved |
| add head-tail balancer (ablation) | COMIC · PLT-COCO | Total mAP: 53.65 -> 55.08 | Improved |
| add positive-negative factor to MFM (on top of H-T) (ablation) | MFM loss · PLT-COCO | Total Shot mAP: 54.44 -> 55.08 | Improved |
| add head-tail factor to MFM (on top of P-N) (ablation) | MFM loss · PLT-COCO | Low Shot mAP: 52.91 -> 55.36 | Improved |
| use dynamic instead of static class distribution for corrected-label weighting (ablation) | RLC · PLT-COCO | Total Shot mAP: 54.67 -> 55.08 | Improved |
| decrease label missing ratio from 50% to 0% (ablation) | COMIC · PLT-COCO | Total Shot mAP: 54.69 -> 57.07 | Improved |
| joint training of head, tail and balanced models versus detached (ablation) | COMIC · PLT-COCO | Total mAP | Mixed |
| treat unannotated labels as negatives with BCE under LT and partial labels | BCE baseline · PLT-COCO | Total Shot mAP: 48.92 vs COMIC 55.08 | Hurt |
| use PL-MLC methods versus LT-MLC methods | baselines · PLT-COCO | mAP | Hurt |
Closest papers by meaning
- MLC-NC: Long-Tailed Multi-Label Image Classification Through the Lens of Neural Collapse (AAAI 2025 · similarity 0.90)
- Long-Tailed Partial Label Learning by Head Classifier and Tail Classifier Cooperation (AAAI 2024 · similarity 0.89)
- Toward Robustness in Multi-Label Classification: A Data Augmentation Strategy against Imbalance and Noise (AAAI 2024 · similarity 0.89)
- Rebalancing Multi-Label Class-Incremental Learning (AAAI 2025 · similarity 0.89)
- Robust Logit Adjustment for Learning with Long-Tailed Noisy Data (AAAI 2025 · similarity 0.87)
- Learning Label Shift Correction for Test-Agnostic Long-Tailed 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.
