Research map / Label-efficient and robust learning / Noisy labels
Transferring Annotator- and Instance-Dependent Transition Matrix for Learning From Crowds
Shikun Li, Xiaobo Xia, Jiankang Deng, Shiming Ge, Tongliang Liu
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 knowledge transfer between neighboring annotators (GCN) over fine-tuning (ablation) | TAIDTM-FT · CIFAR10 (AIDN-40%) | test accuracy: 77.19 -> 79.95 | Improved |
| add knowledge transfer between neighboring annotators (ablation) | TAIDTM-FT · Music | test accuracy: 67.23 -> 69.39 | Improved |
| add global-to-individual transfer (TAIDTM-FT vs BLTM) (ablation) | BLTM · CIFAR10 (AIDN-40%) | test accuracy: 73.06 -> 77.19 | Improved |
| add global-to-individual transfer (TAIDTM-FT vs BLTM) (ablation) | BLTM · LabelMe | test accuracy: 86.16 -> 85.36 | Hurt |
| model annotators individually instead of one source (TAIDTM vs BLTM) | BLTM · CIFAR10 (AIDN-50%) | test accuracy: 60.51 -> 77.18 | Improved |
| TAIDTM vs BLTM on SVHN at high noise | BLTM · SVHN (AIDN-50%) | test accuracy: 80.59 -> 94.48 | Improved |
| TAIDTM vs BLTM on F-MNIST at high noise | BLTM · F-MNIST (AIDN-50%) | test accuracy: 81.10 -> 89.62 | Improved |
| use GCN instead of GAT mapping function (ablation) | TAIDTM · real-world average | test accuracy: 79.52 -> 80.02 | Improved |
| use GCN instead of Deepwalk mapping function (ablation) | TAIDTM · real-world average | test accuracy: 78.83 -> 80.02 | Improved |
| instance-dependent noise-pattern similarity instead of Jaccard, confusion or CBCC similarity (ablation) | TAIDTM · real-world average | test accuracy: 78.93 -> 80.02 | Improved |
| apply Graph Purifying with k=200 (ablation) | TAIDTM · CIFAR10 (AIDN-40%) | test accuracy: 75.33 -> 78.77 | Improved |
| vary number of neighbors k from 1 to 100 (ablation) | TAIDTM · CIFAR10 (AIDN-40%) | test accuracy: 79.90 / 80.16 / 80.19 | No clear effect |
Closest papers by meaning
- Coupled Confusion Correction: Learning from Crowds with Sparse Annotations (AAAI 2024 · similarity 0.90)
- Noisy Label Learning with Instance-Dependent Outliers: Identifiability via Crowd Wisdom (NeurIPS 2024 · similarity 0.88)
- Unbiased Multi-Label Learning from Crowdsourced Annotations (ICML 2024 · similarity 0.88)
- Multi-Instance Multi-Label Classification from Crowdsourced Labels (AAAI 2025 · similarity 0.87)
- Learning Causal Transition Matrix for Instance-dependent Label Noise (AAAI 2025 · similarity 0.87)
- Learning the Latent Causal Structure for Modeling Label Noise (NeurIPS 2024 · 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.
