Research map / Label-efficient and robust learning / Recognition and long-tail
Interactive Multi-Label CNN Learning With Partial Labels
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 |
|---|---|---|---|
| add interactive learning of image and label similarities during end-to-end training | IMCL on Open Images · Open Images test set, all classes | mAP (%): 77.05 -> 78.07 | Helped |
| add image and label similarity graphs, learning both similarities (ablation) | IMCL on Open Images · Open Images test set, Group 5 | mAP (%): 85.49 -> 86.26 | Helped |
| add image similarity learning while fixing label similarities (ablation) | IMCL on Open Images · Open Images test set, Group 5 | mAP (%): 85.49 -> 85.99 | Helped |
| add label similarity learning while fixing image similarities (ablation) | IMCL on Open Images · Open Images test set, Group 5 | mAP (%): 85.49 -> 85.83 | Helped |
| replace learned co-occurrence label graph with learned Wikipedia label similarities (ablation) | IMCL on Open Images · Open Images test set, Group 5 | mAP (%): 86.26 -> 86.32 | No clear effect |
| remove image and label similarity graphs (ablation) | IMCL on Open Images · Open Images test set, Group 5 | mAP (%): 86.26 -> 85.49 | Hurt |
| set label smoothness weight λy from 0 to 1 (ablation) | IMCL on Open Images · Open Images validation set | mAP (%): 78.12 -> 78.44 | Helped |
| increase label smoothness weight λy from 1 to 2 (ablation) | IMCL on Open Images · Open Images validation set | mAP (%): 78.44 -> 78.40 | Hurt |
| add feature smoothness with λf=0.5 instead of λf=0 (ablation) | IMCL on Open Images · Open Images validation set | mAP (%): 78.40 -> 78.44 | Helped |
| increase feature smoothness weight λf from 0.5 to 2 (ablation) | IMCL on Open Images · Open Images validation set | mAP (%): 78.44 -> 78.17 | Hurt |
| vary the maximum number of selected similar images k from 3 to 7 (ablation) | IMCL on Open Images · Open Images validation set | mAP (%): 78.34; 78.44; 78.17 | Mixed |
| use image similarity to favor selecting images from the same class (ablation) | IMCL on CUB · CUB attribute prediction with missing attributes | mAP | Helped |
Closest papers by meaning
- Learning a Deep ConvNet for Multi-Label Classification With Partial Labels (CVPR 2019 · similarity 0.89)
- CNN-RNN: A Unified Framework for Multi-Label Image Classification (CVPR 2016 · similarity 0.88)
- Multi-Label Learning From Single Positive Labels (CVPR 2021 · similarity 0.88)
- ADIOS: Architectures Deep In Output Space (ICML 2016 · similarity 0.87)
- Realistic Evaluation of Deep Partial-Label Learning Algorithms (ICLR 2025 · similarity 0.86)
- Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency (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.