Research map / Learning theory and algorithms / Kernels and regression theory
Model Agnostic Sample Reweighting for Out-of-Distribution Learning
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 |
|---|---|---|---|
| learn sample weights by bilevel optimization using an IRMv1 outer objective | ResNet-18 · CIFARMNIST | Top-1 test accuracy: 51.3 ± 3.0 -> 82.9 \pm 0.4 | Helped |
| learn sample weights by bilevel optimization using an IRMv1 outer objective | ResNet-18 · ColoredObject | Top-1 test accuracy: 87.4\pm 0.6 -> 87.4 ± 0.5 | No clear effect |
| learn sample weights with the validation-set OOD criterion instead of ERM | MAPLE · Waterbirds | Worst-group accuracy: 86.7 -> 91.7 | Helped |
| learn sample weights with the validation-set OOD criterion instead of ERM | MAPLE · CelebA | Worst-group accuracy: 81.1 -> 88.0 | Helped |
| use MAPLE rather than GroupDRO | DRO on Waterbirds · Waterbirds | Worst-group accuracy: 91.4 -> 91.7 | Helped |
| use MAPLE rather than GroupDRO | DRO on CelebA · CelebA | Worst-group accuracy: 88.9 -> 88.0 | Hurt |
| add a sparsity constraint on training sample size (ablation) | MAPLE on Waterbirds · Waterbirds | Inner-loop GPU hours: 8.43 -> 6.74 | Helped |
| transfer sample weights searched on ResNet-18 to ResNet-50 weighted ERM training | MAPLE sample weights on Waterbirds · Waterbirds | Worst-group accuracy: 91.2% -> 91.6\% | Helped |
Closest papers by meaning
- Understanding Why Generalized Reweighting Does Not Improve Over ERM (ICLR 2023 · similarity 0.89)
- Improving Generalization via Meta-Learning on Hard Samples (CVPR 2024 · similarity 0.85)
- Domain Generalisation via Imprecise Learning (ICML 2024 · similarity 0.84)
- Sample-Optimal Agnostic Boosting with Unlabeled Data (ICML 2025 · similarity 0.84)
- Towards Robust Out-of-Distribution Generalization Bounds via Sharpness (ICLR 2024 · similarity 0.84)
- Out-of-distribution Generalization for Total Variation based Invariant Risk Minimization (ICLR 2025 · similarity 0.84)
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.