Research map / Label-efficient and robust learning / Recognition and long-tail
Constrained Submodular Minimization for Missing Labels and Class Imbalance in Multi-label Learning
Baoyuan Wu, Siwei Lyu, Bernard Ghanem
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 class cardinality bounds (MMIB vs MMIB-0) (ablation) | MMIB · Emotions, Scene, Yeast, CAL500, Corel5k | F1-mean: +1.49% to +6.54% across missing-label rates | Improved |
| add class bounds on positive instances (zeta1=1.2, zeta2=3) vs none (ablation) | MMIB · CAL500 0% missing | F1-mean: 32.02 -> 37.35 | Improved |
| add class bounds on positive instances (zeta1=1.2, zeta2=3) vs none (ablation) | MMIB · CAL500 80% missing | F1-mean: 30.24 -> 35.34 | Improved |
| use overly low lower bound zeta1=0.8 with loose upper bound (ablation) | MMIB · CAL500 0% missing | F1-mean: 32.02 -> 31.75 | No clear effect |
| replace discrete st-cut with Lovasz-extension LP (MMIB-0) (ablation) | st-cut · five multi-label datasets with missing labels | F1-mean | Improved |
| use MMIB instead of COCOA for class imbalance | COCOA · Emotions, Scene, Yeast, CAL500 | F1-mean: 64.93 -> 71.07 on Emotions; 29.51 -> 37.78 on CAL500 | Improved |
| MMIB vs COCOA on F1-macro | COCOA · CAL500 | F1-macro: 17.35 -> 16.74 | Hurt |
Closest papers by meaning
- Rebalancing Multi-Label Class-Incremental Learning (AAAI 2025 · similarity 0.89)
- Addressing Imbalance in Multi-Label Classification Using Structured Hellinger Forests (AAAI 2017 · similarity 0.88)
- Toward Robustness in Multi-Label Classification: A Data Augmentation Strategy against Imbalance and Noise (AAAI 2024 · similarity 0.88)
- Generalization Analysis for Multi-Label Learning (ICML 2024 · similarity 0.87)
- Rethinking the Value of Labels for Improving Class-Imbalanced Learning (NeurIPS 2020 · similarity 0.87)
- Addressing Multi-Label Learning with Partial Labels: From Sample Selection to Label Selection (AAAI 2025 · 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.
