Research map / Label-efficient and robust learning / Noisy labels
Noisy Feature Mixup
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 noise to convex combinations of examples or features during mixup | Pre-activated ResNet-18 · CIFAR-10, white noise σ=0.3 | accuracy (%): 71.8 -> 83.0 | Helped |
| Add noise to convex combinations of examples or features during mixup | Pre-activated ResNet-18 · CIFAR-10, salt and pepper γ=0.1 | accuracy (%): 64.9 -> 73.3 | Helped |
| Add noise to convex combinations of examples or features during mixup | Wide ResNet-18 · CIFAR-100 clean | accuracy (%): 80.3 -> 80.9 | Helped |
| Add noise to convex combinations of examples or features during mixup | Wide ResNet-18 · CIFAR-100, white noise σ=0.3 | accuracy (%): 46.7 -> 55.3 | Helped |
| Add noise to convex combinations of examples or features during mixup | Wide ResNet-18 · CIFAR-100, salt and pepper γ=0.1 | accuracy (%): 25.9 -> 34.4 | Helped |
| Add noise to convex combinations of examples or features during mixup | ResNet-50 · ImageNet clean | accuracy (%): 76.7 -> 77.0 | Helped |
| Add noise to convex combinations of examples or features during mixup | ResNet-50 · ImageNet, white noise σ=0.5 | accuracy (%): 57.5 -> 60.1 | Helped |
| Use Noisy Feature Mixup instead of Manifold Mixup | ResNet-50 · ImageNet, salt and pepper γ=0.15 | accuracy (%): 49.5 -> 54.4 | Helped |
| Increase NFM mixing parameter α from 0.2 to 1.0 (ablation) | ResNet-50 · ImageNet, salt and pepper γ=0.15 | accuracy (%): 52.3 -> 54.4 | Helped |
| Increase NFM mixing parameter α from 0.2 to 1.0 (ablation) | ResNet-50 · ImageNet, white noise σ=0.5 | accuracy (%): 60.1 -> 60.0 | No clear effect |
Closest papers by meaning
- Using Mixup as a Regularizer Can Surprisingly Improve Accuracy & Out-of-Distribution Robustness (NeurIPS 2022 · similarity 0.89)
- C-Mixup: Improving Generalization in Regression (NeurIPS 2022 · similarity 0.88)
- A Unified Analysis of Mixed Sample Data Augmentation: A Loss Function Perspective (NeurIPS 2022 · similarity 0.88)
- Catch-Up Mix: Catch-Up Class for Struggling Filters in CNN (AAAI 2024 · similarity 0.87)
- Robust Image Denoising through Adversarial Frequency Mixup (CVPR 2024 · similarity 0.86)
- Adversarial AutoMixup (ICLR 2024 · similarity 0.85)
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.