Research map / Label-efficient and robust learning / Recognition and long-tail
Dual Graph Disambiguation for Multi-Instance Partial-Label Learning
Zhen Zhu, Kai Tang, Songhe Feng, Yixuan Tang, Haobo Wang, Gengyu Lyu, Cheng Peng, Yining Sun
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 sharpening regularization Lsp (ablation) | DualG · MNIST-MIPL (r=1) | accuracy: 0.892 -> 0.993 | Improved |
| add graph loss Lgraph (ablation) | DualG · MNIST-MIPL (r=1) | accuracy: 0.869 -> 0.993 | Improved |
| add bag-level attention loss (L1 feature) (ablation) | DualG · FMNIST-MIPL (r=2) | accuracy: 0.745 -> 0.810 | Improved |
| add instance-level fusion loss (L2 feature) (ablation) | DualG · FMNIST-MIPL (r=1) | accuracy: 0.824 -> 0.901 | Improved |
| add Lsp / Lgraph under high ambiguity (ablation) | DualG · FMNIST-MIPL (r=3) | accuracy: 0.612 -> 0.624 | No clear effect |
| add Lgraph at moderate ambiguity (ablation) | DualG · FMNIST-MIPL (r=2) | accuracy: 0.808 -> 0.810 | No clear effect |
| add dual graph components (bag graph, instance graph, Sclean filter) (ablation) | DualG · MNIST/FMNIST | accuracy | Improved |
| use DualG instead of MIPLGP | DualG · FMNIST-MIPL (r=3) | accuracy: 0.670 -> 0.624 | Hurt |
| use DualG instead of DEMIPL | DualG · MNIST-MIPL (r=3) | accuracy: 0.677 -> 0.811 | Improved |
| use DualG instead of MIPLGP | DualG · SIVAL-MIPL (r=3) | accuracy: 0.422 -> 0.455 | Improved |
| use DualG instead of MIPLGP | DualG · Birdsong-MIPL (r=2) | accuracy: 0.672 -> 0.659 | Hurt |
| use DualG instead of MIPLGP | DualG · CRC-MIPL (Row features) | accuracy: 0.432 -> 0.420 | Hurt |
Closest papers by meaning
- Adaptive Graph Guided Disambiguation for Partial Label Learning (TPAMI 2021 · similarity 0.91)
- Distilling Reliable Knowledge for Instance-Dependent Partial Label Learning (AAAI 2024 · similarity 0.89)
- Disentangled Partial Label Learning (AAAI 2024 · similarity 0.87)
- Navigating Real-World Partial Label Learning: Unveiling Fine-Grained Images with Attributes (AAAI 2024 · similarity 0.87)
- Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency (ICML 2024 · similarity 0.87)
- Fast Multi-Instance Partial-Label Learning (AAAI 2025 · 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.
