Research map / Label-efficient and robust learning / Recognition and long-tail
On Mixup Training: Improved Calibration and Predictive Uncertainty for Deep Neural Networks
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 |
|---|---|---|---|
| train with mixup | ResNet-18 · CIFAR-10 | ECE: 0.023 -> 0.019 | Helped |
| train with mixup | ResNet-18 · CIFAR-100 | ECE: 0.049 -> 0.035 | Helped |
| train with mixup | ResNet-18 · CIFAR-10 | test accuracy: 95.12 -> 96.16 | Helped |
| increase mixup alpha to 1.0 (ablation) | Mixup · CIFAR-10 ResNet-18 | ECE: 0.019 -> 0.1 | Hurt |
| mix features only, keeping hard labels (remove label mixing) (ablation) | Mixup · CIFAR / STL-10 benchmarks | ECE | Hurt |
| train with mixup | VGG-16 · STL-10/Gaussian noise OOD | AUROC: 73.28 -> 95.93 | Helped |
| apply temperature scaling | VGG-16 baseline · STL-10/ImageNet OOD | AUROC: 80.57 -> 56.2 | Hurt |
| mixup vs label smoothing on ResNet-50 | ImageNet · ImageNet | ECE | No clear effect |
| train with mixup | ImageNet ResNets · ImageNet | accuracy: ≈ 0.4 | Helped |
| use manifold mixup instead of mixup (200 epochs) | PreActResNet-18 · CIFAR-100 | ECE: 0.047 -> 0.22 | Hurt |
| train with mixup on embeddings | Sentence CNN · MR, TREC, IMDB | ECE | Helped |
| extended training to 1000 epochs | Mixup · CIFAR-10/CIFAR-100 | ECE | No clear effect |
Closest papers by meaning
- On the Pitfall of Mixup for Uncertainty Calibration (CVPR 2023 · similarity 0.90)
- Pushing Boundaries: Mixup's Influence on Neural Collapse (ICLR 2024 · similarity 0.88)
- Catch-Up Mix: Catch-Up Class for Struggling Filters in CNN (AAAI 2024 · similarity 0.88)
- Tailoring Mixup to Data for Calibration (ICLR 2025 · similarity 0.88)
- On the Limitations of Temperature Scaling for Distributions with Overlaps (ICLR 2024 · similarity 0.87)
- Adversarial AutoMixup (ICLR 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.