Research map / Label-efficient and robust learning / Recognition and long-tail
Mitigating Instance Entanglement in Instance-Dependent Partial Label Learning
Rui Zhao, Bin Shi, Kai Sun, Bo Dong
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 representation learning with class-specific augmentation (ablation) | CAD · CIFAR-10 (ID-PLL) | accuracy: 91.21 -> 93.57 | Improved |
| add confidence adjustment (weighted penalty) loss (ablation) | CAD · CIFAR-10 (ID-PLL) | accuracy: 93.32 -> 93.57 | No clear effect |
| add confidence adjustment loss (ablation) | CAD · Fashion-MNIST (ID-PLL) | accuracy: 91.19 -> 92.14 | Improved |
| add representation learning (ablation) | CAD · Fashion-MNIST (ID-PLL) | accuracy: 91.48 -> 92.14 | Improved |
| add both modules to vanilla weighted-loss distillation (ablation) | CAD · CIFAR-10 (ID-PLL) | accuracy: 87.81 -> 93.57 | Improved |
| diffusion-edited augmentation instead of CAM-based (ablation) | CAD · Flower and Oxford-IIIT Pet | accuracy: Pet 74.56 -> 69.46; Flower 49.67 -> 47.88 | Hurt |
| diffusion-edited augmentation instead of CAM-based (ablation) | CAD · CIFAR-100 (ID-PLL) | accuracy: 69.08 -> 72.03 | Improved |
| use detailed fine-grained class descriptions in edit prompts (ablation) | CAD · Oxford-IIIT Pet (ID-PLL) | accuracy: 69.46 -> 76.23 | Improved |
| CAD versus DIRK on top 0.001% entangled pairs | CAD · CIFAR-10 (ID-PLL) | accuracy: 74.09 -> 83.37 | Improved |
| add diffusion-edited instances directly to ABLE/DIRK as supervision (ablation) | DIRK/ABLE · ABLE, DIRK | accuracy | Hurt |
| CAD overall versus DIRK | CAD · CIFAR-10 (ID-PLL) | accuracy: 90.87 -> 93.57 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- PICO — PiCO: Contrastive Label Disambiguation for Partial Label Learning (ICLR 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Disentangled Partial Label Learning (AAAI 2024 · similarity 0.89)
- Distilling Reliable Knowledge for Instance-Dependent Partial Label Learning (AAAI 2024 · similarity 0.88)
- Reduction-based Pseudo-label Generation for Instance-dependent Partial Label Learning (NeurIPS 2025 · similarity 0.85)
- Partial Label Learning with a Partner (AAAI 2024 · similarity 0.85)
- Partial Label Causal Representation Learning for Instance-Dependent Supervision and Domain Generalization (AAAI 2025 · similarity 0.85)
- Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency (ICML 2024 · similarity 0.84)
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.
