Research map / Learning theory and algorithms / Prediction and decision losses
Principled learning method for Wasserstein distributionally robust optimization with local perturbations
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 WDRO or WDRO+MIX instead of ERM or MIXUP | CIFAR-10 and CIFAR-100 classifiers · 1% salt-and-pepper test; all reported training sizes | accuracy | Helped |
| Use WDRO or WDRO+MIX instead of ERM or MIXUP | CIFAR-10 and CIFAR-100 classifiers · Clean test; larger training sizes | accuracy | Hurt |
| Add Mixup to WDRO | WDRO · CIFAR-10 clean test, 2500 training images | accuracy: 77.1±0.7 -> 80.8±0.7 | No clear effect |
| Add the WDRO gradient penalty | MIXUP · CIFAR-10 1% salt-and-pepper test, 5000 training images | accuracy: 76.4±1.7 -> 79.6±0.9 | Helped |
| Use WDRO instead of ERM | CIFAR-10 classifier · CIFAR-10 accuracy reduction, 2% noisy pixels | accuracy reduction: 21.1±1.9 -> 13.2± 0.5 | Helped |
| Add WDRO to Mixup | MIXUP · CIFAR-10 accuracy reduction, 2% noisy pixels | accuracy reduction: 24.3±1.4 -> 12.7±0.8 | Helped |
| Use WDRO instead of ERM | CIFAR-100 classifier · CIFAR-100 accuracy reduction, 2% noisy pixels | accuracy reduction: 42.8±2.3 -> 26.5±1.0 | Helped |
| Add WDRO to Mixup | MIXUP · CIFAR-100 accuracy reduction, 2% noisy pixels | accuracy reduction: 45.9±3.4 -> 29.7±0.7 | Helped |
| Add the gradient penalty | ERM or MIXUP · CIFAR-10 test images; gradient norm | gradient norm | Helped |
Closest papers by meaning
- Outlier-Robust Wasserstein DRO (NeurIPS 2023 · similarity 0.93)
- Knowledge-Guided Wasserstein Distributionally Robust Optimization (ICML 2025 · similarity 0.92)
- Provable Robust Overfitting Mitigation in Wasserstein Distributionally Robust Optimization (ICLR 2025 · similarity 0.91)
- Universal generalization guarantees for Wasserstein distributionally robust models (ICLR 2025 · similarity 0.88)
- Wasserstein Distributionally Robust Optimization through the Lens of Structural Causal Models and Individual Fairness (NeurIPS 2024 · similarity 0.87)
- Group Distributionally Robust Optimization with Flexible Sample Queries (NeurIPS 2026 · 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.