Research map / Label-efficient and robust learning / Recognition and long-tail
Learning with Unsure Responses
Kunihiro Takeoka, Yuyang Dong, Masafumi Oyamada
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 unsure loss (absolute form) as regularizer | logistic regression · MNIST synthetic single annotator | F1: 0.85 -> 0.88 | Improved |
| add unsure loss (absolute form) | logistic regression · MNIST synthetic single annotator | AUC: 0.76 -> 0.80 | Improved |
| add unsure loss (log form) | SVM · MNIST synthetic single annotator | F1: 0.85 -> 0.84 | No clear effect |
| add unsure loss (absolute form) | SVM · MNIST synthetic single annotator | F1: 0.85 -> 0.87 | Improved |
| add unsure loss (squared form) to personal classifiers | PC · MNIST synthetic DS annotators | F1: 0.82 ± 0.10 -> 0.93 ± 0.03 | Improved |
| add unsure loss (log form) to personal classifiers | PC · MNIST synthetic PC annotators | F1: 0.78 ± 0.20 -> 0.83 ± 0.13 | Improved |
| add unsure loss (abs form) | logistic regression · DOG-WOLF single annotator | F1: 0.80±0.05 -> 0.85±0.04 | Improved |
| add unsure loss (log form) | logistic regression · DOG-WOLF single annotator | AUC: 0.81±0.01 -> 0.81±0.01 | No clear effect |
| add unsure loss (log form) | SVM · DOG-WOLF single annotator | F1: 0.74 ±0.06 -> 0.74 ±0.06 | No clear effect |
| add unsure loss to personal classifiers | PC · DOG-WOLF five annotators | F1: 0.60±0.08 -> 0.63±0.06 | Improved |
| add unsure loss to personal classifiers | PC · DOG-WOLF five annotators | AUC: 0.50±0.07 -> 0.51±0.06 | No clear effect |
| treat unsure instances as unlabeled (semi-supervised) rather than via unsure loss | PC · MNIST synthetic DS annotators | F1: 0.82 ± 0.10 -> 0.63 ± 0.07 | Hurt |
Closest papers by meaning
- Learning with Noisy Labels Revisited: A Study Using Real-World Human Annotations (ICLR 2022 · similarity 0.84)
- Collaborative Refining for Learning from Inaccurate Labels (NeurIPS 2024 · similarity 0.84)
- Noisy Label Learning with Instance-Dependent Outliers: Identifiability via Crowd Wisdom (NeurIPS 2024 · similarity 0.84)
- TrainRef: Curating Data with Label Distribution and Minimal Reference for Accurate Prediction and Reliable Confidence (ICLR 2026 · similarity 0.84)
- Coupled Confusion Correction: Learning from Crowds with Sparse Annotations (AAAI 2024 · similarity 0.83)
- Enhanced Sample Selection with Confidence Tracking: Identifying Correctly Labeled Yet Hard-to-Learn Samples in Noisy Data (AAAI 2025 · similarity 0.83)
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.
