Research map / Label-efficient and robust learning / Noisy labels
Learning Noise Transition Matrix from Only Noisy Labels via Total Variation Regularization
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 TVD instead of Dual-T | Noisy-label classification on Clothing1M · Clothing1M | accuracy: 70.67 -> 71.65 | Helped |
| Use total variation regularization with gradient-based transition estimation instead of Dual-T | ResNet-18 · CIFAR-10 Pair | accuracy: 80.86(1.03) -> 89.78(0.16) | Helped |
| Use total variation regularization with Dirichlet posterior update instead of Dual-T | ResNet-18 · CIFAR-100 Pair | accuracy: 46.22(0.72) -> 56.02(0.82) | Helped |
| Use gradient-based transition estimation instead of Dirichlet posterior update | Total variation regularization · CIFAR-10 Pair | accuracy: 88.47(0.29) -> 89.78(0.16) | Helped |
| Use Dirichlet posterior update instead of gradient-based transition estimation | Total variation regularization · CIFAR-100 Pair | accuracy: 45.68(0.85) -> 56.02(0.82) | Helped |
| Use gradient-based transition estimation instead of Dirichlet posterior update | Total variation regularization · CIFAR-100 Symm. | accuracy: 52.54(0.45) -> 57.33(0.42) | Helped |
| Use Dirichlet posterior update instead of gradient-based transition estimation | Total variation regularization · CIFAR-100 Pair² | accuracy: 44.38(0.72) -> 49.18(0.53) | Helped |
| Use gradient-based transition estimation instead of Dirichlet posterior update | Total variation regularization · CIFAR-10 Symm. | accuracy: 83.03(0.24) -> 82.60(0.38) | No clear effect |
Closest papers by meaning
- Dual T: Reducing Estimation Error for Transition Matrix in Label-noise Learning (NeurIPS 2020 · similarity 0.92)
- Estimation of the Label-Noise Transition Matrix with Performance Guarantees via Selective Classification (NeurIPS 2026 · similarity 0.92)
- Estimating Noisy Class Posterior with Part-level Labels for Noisy Label Learning (CVPR 2024 · similarity 0.89)
- Learning the Latent Causal Structure for Modeling Label Noise (NeurIPS 2024 · similarity 0.89)
- Learning Causal Transition Matrix for Instance-dependent Label Noise (AAAI 2025 · similarity 0.87)
- Label Noise: Ignorance Is Bliss (NeurIPS 2024 · 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.