Research map / Label-efficient and robust learning / Knowledge distillation
Knowledge Amalgamation for Multi-Label Classification via Label Dependency Transfer
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 |
|---|---|---|---|
| Use ANT to amalgamate multi-label teachers instead of the strongest F1 Micro baseline | Multi-label knowledge amalgamation · EMOTIONS | F1 Micro: .4372±.0815 -> .4545±.0398 | Helped |
| Use ANT to amalgamate multi-label teachers instead of the strongest F1 Micro baseline | Multi-label knowledge amalgamation · SCENE | F1 Micro: .4645±.0621 -> .4840±.0398 | Helped |
| Use ANT to amalgamate multi-label teachers instead of the strongest F1 Micro baseline | Multi-label knowledge amalgamation · YELP | F1 Micro: .4151±.0121 -> .4217±.0113 | Helped |
| Use ANT to amalgamate multi-label teachers instead of the strongest F1 Micro baseline | Multi-label knowledge amalgamation · YEAST | F1 Micro: .4917±.0050 -> .5073±.0061 | Helped |
| Use ANT to amalgamate multi-label teachers instead of the strongest F1 Micro baseline | Multi-label knowledge amalgamation · BIRD | F1 Micro: .0937±.0295 -> .0132±.0263 | Hurt |
| Use ANT to amalgamate multi-label teachers instead of the strongest F1 Micro baseline | Multi-label knowledge amalgamation · TMC | F1 Micro: .1967±.0023 -> .0825±.0033 | Hurt |
| Use ANT to amalgamate multi-label teachers instead of the strongest F1 Micro baseline | Multi-label knowledge amalgamation · GENBASE | F1 Micro: .0734±.0265 -> .0740±.0266 | Helped |
| Use ANT to amalgamate multi-label teachers instead of the strongest F1 Micro baseline | Multi-label knowledge amalgamation · MEDICAL | F1 Micro: .0114±.0146 -> .0119±.0151 | Helped |
| Reduce the proportion of labels shared between teachers from 50% to 25% (ablation) | ANT on SCENE · SCENE | F1 Micro: .305±.009 -> .484±.040 | Helped |
| Reduce the proportion of labels shared between teachers from 25% to 0% (ablation) | ANT on SCENE · SCENE | F1 Micro: .484±.040 -> .685±.032 | Helped |
| Increase the number of teachers from 3 to 4 (ablation) | ANT on YELP · YELP | F1 Micro: .488±.010 -> .373±.003 | Hurt |
| Increase the number of teachers from 4 to 5 (ablation) | ANT on YELP · YELP | F1 Micro: .373±.003 -> .382±.002 | Helped |
Closest papers by meaning
- Multi-Label Iterated Learning for Image Classification With Label Ambiguity (CVPR 2022 · similarity 0.89)
- Distilling Reliable Knowledge for Instance-Dependent Partial Label Learning (AAAI 2024 · similarity 0.88)
- Specifying What You Know or Not for Multi-Label Class-Incremental Learning (AAAI 2025 · similarity 0.87)
- Rebalancing Multi-Label Class-Incremental Learning (AAAI 2025 · similarity 0.87)
- Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency (ICML 2024 · similarity 0.87)
- Enhancing Multi-Label Classification via Dynamic Label-Order Learning (AAAI 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.