Research map / Label-efficient and robust learning / Recognition and long-tail
Multi-Label Classification With Partial Annotations Using Class-Aware Selective Loss
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 |
|---|---|---|---|
| Use class-aware Selective mode with both likelihood and prior conditions instead of Negative mode | P-ASL · OpenImages V6 | mAP(C): 86.28 -> 86.72 | Helped |
| Use class-aware Selective mode with both likelihood and prior conditions instead of Negative mode | P-ASL · OpenImages V6 | mAP(O): 92.34 -> 93.57 | Helped |
| Add the likelihood condition to prior-only selection (ablation) | P-ASL Selective · OpenImages V6 | mAP(C): 86.46 -> 86.72 | Helped |
| Add the likelihood condition to prior-only selection (ablation) | P-ASL Selective · OpenImages V6 | mAP(O): 93.27 -> 93.57 | Helped |
| Add the prior condition to likelihood-only selection (ablation) | P-ASL Selective · OpenImages V6 | mAP(C): 86.36 -> 86.72 | Helped |
| Add the prior condition to likelihood-only selection (ablation) | P-ASL Selective · OpenImages V6 | mAP(O): 93.25 -> 93.57 | Helped |
| Use hard prior selection instead of soft prior weighting (ablation) | P-ASL Selective · OpenImages V6 | mAP(C): 86.62 -> 86.72 | Helped |
| Use hard prior selection instead of soft prior weighting (ablation) | P-ASL Selective · OpenImages V6 | mAP(O): 93.59 -> 93.57 | No clear effect |
| Lower the annotated-negative focusing parameter while keeping the un-annotated focusing parameter at 7 (ablation) | P-ASL · OpenImages V6 | mAP | Helped |
| Estimate class priors with an Ignore-trained model rather than using Negative mode | Class-distribution estimator · MS-COCO fixed per class | Spearman correlation: 0.81 | Helped |
| Train with Ignore rather than Negative mode under random-per-annotation partial labels | Multi-label classification · MS-COCO RPA | mAP | Helped |
| Train with Negative rather than Ignore mode under fixed-per-class partial labels | Multi-label classification · MS-COCO FPC | mAP | Helped |
Closest papers by meaning
- Bridging the Gap Between Model Explanations in Partially Annotated Multi-Label Classification (CVPR 2023 · similarity 0.91)
- Rethinking Consistent Multi-Label Classification under Inexact Supervision (ICLR 2026 · similarity 0.90)
- Large Loss Matters in Weakly Supervised Multi-Label Classification (CVPR 2022 · similarity 0.90)
- Partial-Label Regression (AAAI 2023 · similarity 0.89)
- Addressing Multi-Label Learning with Partial Labels: From Sample Selection to Label Selection (AAAI 2025 · similarity 0.88)
- Double-Layer Hybrid-Label Identification Feature Selection for Multi-View Multi-Label Learning (AAAI 2024 · similarity 0.88)
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.