Research map / Label-efficient and robust learning / Noisy labels
Training Noise-Robust Deep Neural Networks via Meta-Learning
Check 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 |
|---|---|---|---|
| Learn the transition matrix directly using meta-learning and clean validation loss instead of estimating it heuristically | MLC · MNIST, mean across noise types and ratios | accuracy: 98.34 -> 98.64 | Helped |
| Learn the transition matrix directly using meta-learning and clean validation loss instead of estimating it heuristically | MLC · CIFAR-10, mean across noise types and ratios | accuracy: 83.36 -> 83.84 | Helped |
| Learn the transition matrix directly using meta-learning and clean validation loss instead of estimating it heuristically | MLC · CIFAR-100, mean across Flip-Random and Flip-To-One noise ratios | accuracy: 55.15 -> 55.53 | Helped |
| Learn the transition matrix directly using meta-learning and clean validation loss instead of estimating it heuristically | MLC · Twitter, mean across noise types and ratios | accuracy: 85.88 -> 86.35 | Helped |
| Use MLC rather than Mask | Clothing1M · Clothing1M test | accuracy: 71.10 vs. 71.06 | No clear effect |
| Increase the number of clean validation images (ablation) | MLC · CIFAR-10, 30% noise | accuracy | Helped |
| Increase the clean validation set beyond 100 images (ablation) | MLC · CIFAR-10, 30% noise | accuracy | No clear effect |
| Add a fully connected noise-correction layer and alternate training with clean validation data instead of using MLC meta-learning (ablation) | Baseline (FC) · CIFAR-10, various noise types | accuracy | Helped |
| Use MLC meta-learning rather than the fully connected noise-correction-layer baseline (ablation) | CIFAR-10 noise correction · CIFAR-10, various noise types | accuracy | Helped |
Closest papers by meaning
- Learning to Learn From Noisy Labeled Data (CVPR 2019 · similarity 0.91)
- $\epsilon$-Softmax: Approximating One-Hot Vectors for Mitigating Label Noise (NeurIPS 2024 · similarity 0.88)
- Enhancing Noise-Robust Losses for Large-Scale Noisy Data Learning (AAAI 2025 · similarity 0.87)
- Learning from Noisy Labels via Self-Taught On-the-Fly Meta Loss Rescaling (AAAI 2025 · similarity 0.87)
- Handling Label Noise via Instance-Level Difficulty Modeling and Dynamic Optimization (NeurIPS 2025 · similarity 0.87)
- Mitigating Label Noise through Data Ambiguation (AAAI 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.