Research map / Label-efficient and robust learning / Recognition and long-tail
IGNORE: Information Gap-based False Negative Loss Rejection for Single Positive Multi-Label Learning
Gyeong Ryeol Song, Noo-ri Kim, Jin-Seop Lee, Jee-Hyong Lee
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 IGNORE rejection, SPBC and consistency regularization to AN loss (ablation) | Q2L · PASCAL VOC 2012 | mAP: 87.9 -> 90.1 | Improved |
| add logit average threshold to IGNORE+SPBC+consistency (ablation) | Q2L · NUSWIDE | mAP: 52.0 -> 52.8 | Improved |
| add logit average threshold to IGNORE+SPBC+consistency (ablation) | Q2L · CUB | mAP: 22.8 -> 23.4 | Improved |
| add logit average threshold (ablation) | Q2L · PASCAL VOC 2012 | mAP: 90.0 -> 90.1 | No clear effect |
| apply consistency regularization alone without IGNORE and SPBC (ablation) | Q2L · NUSWIDE and CUB | mAP: NUSWIDE 49.2 -> 47.5; CUB 19.7 -> 11.1 | Hurt |
| apply consistency regularization alone (ablation) | Q2L · COCO | mAP: 73.5 -> 73.6 | No clear effect |
| add IGNORE and SPBC on top of consistency regularization (ablation) | Q2L · COCO | mAP: 73.6 -> 76.2 | Improved |
| add logit average threshold to gap-only rejection (ablation) | IGNORE · PASCAL VOC 2012 | rejection ratio FN/TN | Improved |
| vary CAM threshold used to build masked image (ablation) | IGNORE · SPML datasets | test mAP | No clear effect |
| use IGNORE versus best prior method with ResNet-50 | ResNet-50 on COCO · MS COCO (SPML) | mAP: 72.89 -> 72.00 | Hurt |
Later papers that compared with this method
1 later accepted paper reports Ignore as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- Regretful Decisions under Label Noise (ICLR 2025 · better on 1 of 2 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Acknowledging the Unknown for Multi-Label Learning with Single Positive Labels (ECCV 2022 · similarity 0.91)
- More Reliable Pseudo-labels, Better Performance: A Generalized Approach to Single Positive Multi-label Learning (ICCV 2025 · similarity 0.89)
- Can Class-Priors Help Single-Positive Multi-Label Learning? (NeurIPS 2025 · similarity 0.88)
- Semi-supervised Multi-label Learning with Balanced Binary Angular Margin Loss (NeurIPS 2024 · similarity 0.85)
- Addressing Multi-Label Learning with Partial Labels: From Sample Selection to Label Selection (AAAI 2025 · similarity 0.85)
- Enhanced Sample Selection with Confidence Tracking: Identifying Correctly Labeled Yet Hard-to-Learn Samples in Noisy Data (AAAI 2025 · similarity 0.85)
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.
