Research map / Label-efficient and robust learning / Recognition and long-tail
On the Robustness of Average Losses for Partial-Label Learning
Jiaqi Lv, Biao Liu, Lei Feng, Ning Xu, Miao Xu, Bo An, Gang Niu, Xin Geng, Masashi Sugiyama
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 |
|---|---|---|---|
| use bounded MAE loss instead of unbounded CCE | ABS partial-label learning · CIFAR-10 ConvNet case 4 | test accuracy (%): 17.29 -> 51.96 | Improved |
| use bounded MAE loss instead of CCE | ABS · CIFAR-10 MLP case 1 | test accuracy (%): 22.47 -> 41.46 | Improved |
| use MAE instead of CCE | ABS · Fashion-MNIST MLP case 4 | test accuracy (%): 25.76 -> 82.41 | Improved |
| use MAE instead of CCE | ABS · MNIST Linear case 4 | test accuracy (%): 67.78 -> 90.37 | Improved |
| use GCE instead of MAE | ABS · CIFAR-10 ConvNet case 4 | test accuracy (%): 51.96 -> 43.83 | Hurt |
| use GCE instead of MAE | ABS · CIFAR-10 ConvNet case 3 | test accuracy (%): 68.88 -> 71.65 | Improved |
| use MAE instead of GCE on a linear model | ABS · Fashion-MNIST Linear case 1 | test accuracy (%): 81.50 -> 82.50 | No clear effect |
| add robust warm start (MAE for 20 epochs) to PRODEN | PRODEN · CIFAR-10 ConvNet case 3 | test accuracy (%): 64.59 -> 66.86 | Improved |
| add robust warm start to PRODEN with early stopping | PRODEN · CIFAR-10 ConvNet case 3 | test accuracy (%): 73.05 -> 75.99 | Improved |
| add robust warm start to PRODEN | PRODEN · CIFAR-100 ConvNet case 3 | test accuracy (%): 34.78 -> 36.27 | Improved |
| add robust warm start and early stopping to PRODEN | PRODEN · CIFAR-100 ConvNet case 4 | test accuracy (%): 37.37 -> 47.86 | Improved |
| add early stopping to PRODEN | PRODEN · CIFAR-100 ConvNet case 4 | test accuracy (%): 37.37 -> 47.29 | Improved |
Closest papers by meaning
- Progressive Identification of True Labels for Partial-Label Learning (ICML 2020 · similarity 0.91)
- Realistic Evaluation of Deep Partial-Label Learning Algorithms (ICLR 2025 · similarity 0.90)
- What Makes Partial-Label Learning Algorithms Effective? (NeurIPS 2024 · similarity 0.90)
- Conformal Prediction for Partial Label Learning (AAAI 2025 · similarity 0.89)
- Forming Auxiliary High-confident Instance-level Loss to Promote Learning from Label Proportions (CVPR 2025 · similarity 0.88)
- CroSel: Cross Selection of Confident Pseudo Labels for Partial-Label Learning (CVPR 2024 · similarity 0.88)
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.
