Research map / Label-efficient and robust learning / Noisy labels
Learning With Noisy Labels via Self-Supervised Adversarial Noisy Masking
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 |
|---|---|---|---|
| Add SANM to DivideMix | DivideMix · CIFAR-10 symmetric 90% noise | accuracy: 76.0 -> 92.3 | Helped |
| Add SANM to DivideMix | DivideMix · CIFAR-100 symmetric 90% noise | accuracy: 31.5 -> 43.5 | Helped |
| Add SANM to C2D | C2D · CIFAR-100 symmetric 90% noise | accuracy: 58.7 -> 61.9 | Helped |
| Add SANM to DivideMix | DivideMix · CIFAR-10 asymmetric noise, 20% and 40% | accuracy: 93.4 -> 95.4; 93.4 -> 94.8 | Helped |
| Add SANM to DivideMix | DivideMix · Clothing-1M | accuracy: 74.76 -> 75.63 | Helped |
| Add SANM to vanilla CE | CE · CIFAR-100 symmetric 80% noise | accuracy: 19.9 -> 34.8 | Helped |
| Add AMG adversarial noisy masking (ablation) | DivideMix · CIFAR-10 symmetric 90% noise | accuracy: 76.0 -> 91.4 | Helped |
| Add AMG adversarial noisy masking (ablation) | DivideMix · CIFAR-100 symmetric 90% noise | accuracy: 31.5 -> 42.7 | Helped |
| Add NLR noisy-label regularization to AMG (ablation) | SANM · CIFAR-10 symmetric 90% noise | accuracy: 91.4 -> 91.6 | Helped |
| Add NLR noisy-label regularization to AMG (ablation) | SANM · CIFAR-100 symmetric 90% noise | accuracy: 42.7 -> 43.0 | Helped |
| Add SMR reconstruction to AMG+NLR (ablation) | SANM · CIFAR-10 symmetric 90% noise | accuracy: 91.6 -> 92.3 | Helped |
| Add SMR reconstruction to AMG+NLR (ablation) | SANM · CIFAR-100 symmetric 90% noise | accuracy: 43.0 -> 43.5 | Helped |
Closest papers by meaning
- Label Noise in Adversarial Training: A Novel Perspective to Study Robust Overfitting (NeurIPS 2022 · similarity 0.86)
- Toward Robustness against Label Noise in Training Deep Discriminative Neural Networks (NeurIPS 2017 · similarity 0.86)
- SELF: Learning to Filter Noisy Labels with Self-Ensembling (ICLR 2020 · similarity 0.84)
- Adversarial Label Learning (AAAI 2019 · similarity 0.84)
- Soften to Defend: Towards Adversarial Robustness via Self-Guided Label Refinement (CVPR 2024 · similarity 0.83)
- Self-cognitive Denoising in the Presence of Multiple Noisy Label Sources (ICML 2024 · similarity 0.82)
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.