Research map / Label-efficient and robust learning / Recognition and long-tail
Bridging the Gap Between Model Explanations in Partially Annotated Multi-Label Classification
Read the paperCheck 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 |
|---|---|---|---|
| Add BoostLU at inference to AN (ablation) | AN multi-label classifier · PASCAL VOC and MS COCO | mAP: 86.10 -> 87.31; 64.58 -> 66.27 | Helped |
| Apply BoostLU during training as well as inference without filtering false negatives (ablation) | AN multi-label classifier · PASCAL VOC and MS COCO | mAP: 87.31 -> 86.73; 66.27 -> 65.33 | Mixed |
| Add LL-R to AN (ablation) | AN multi-label classifier · PASCAL VOC and MS COCO | mAP: 86.10 -> 88.24; 64.58 -> 70.60 | Helped |
| Combine LL-R with BoostLU during training and inference (ablation) | LL-R · PASCAL VOC and MS COCO | mAP: 88.24 -> 89.27; 70.60 -> 72.82 | Helped |
| Apply BoostLU during inference with LL-R, but omit it during training (ablation) | LL-R · PASCAL VOC and MS COCO | mAP: 88.24 -> 88.90; 70.60 -> 70.87 | Helped |
| Apply BoostLU during training with LL-R, but omit it during inference (ablation) | LL-R · PASCAL VOC and MS COCO | mAP: 88.24 -> 87.18; 70.60 -> 68.45 | Hurt |
| Combine BoostLU with LL-R | LL-R · PASCAL VOC, MS COCO, NUSWIDE, and CUB | mAP: 88.27 -> 89.29; 70.70 -> 72.89; 48.76 -> 49.59; 19.56 -> 19.80 | Helped |
| Combine BoostLU with LL-Ct | LL-Ct · PASCAL VOC, MS COCO, NUSWIDE, and CUB | mAP: 87.79 -> 88.61; 70.29 -> 71.78; 48.08 -> 48.37; 19.06 -> 19.25 | Helped |
| Combine BoostLU with LL-Cp | LL-Cp · PASCAL VOC, MS COCO, NUSWIDE, and CUB | mAP: 87.44 -> 87.81; 70.27 -> 71.41; 47.92 -> 48.61; 19.21 -> 19.34 | Helped |
| Combine BoostLU with LL-R | LL-R · OpenImages V3, all classes | mAP: 82.53 -> 83.86 | Helped |
| Combine BoostLU with LL-Ct | LL-Ct · OpenImages V3, all classes | mAP: 82.58 -> 83.94 | Helped |
| Combine BoostLU with LL-Cp | LL-Cp · OpenImages V3, all classes | mAP: 82.46 -> 84.04 | Helped |
Closest papers by meaning
- Multi-Label Classification With Partial Annotations Using Class-Aware Selective Loss (CVPR 2022 · similarity 0.91)
- Multi-Label Learning From Single Positive Labels (CVPR 2021 · similarity 0.90)
- Rethinking Consistent Multi-Label Classification under Inexact Supervision (ICLR 2026 · similarity 0.89)
- Addressing Multi-Label Learning with Partial Labels: From Sample Selection to Label Selection (AAAI 2025 · similarity 0.88)
- Adaptive Integration of Partial Label Learning and Negative Learning for Enhanced Noisy Label Learning (AAAI 2024 · similarity 0.86)
- Counterfactual Reasoning for Multi-Label Image Classification via Patching-Based Training (ICML 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.