Research map / Label-efficient and robust learning / Domain adaptation and generalization
Gradient Matching for Domain Generalization
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 Fish instead of ERM | domain generalization models · Camelyon17-Wilds test | average accuracy (%): 70.3 -> 74.7 | Helped |
| Use Fish instead of ERM | domain generalization models · FMoW-Wilds test | worst-group accuracy (%): 32.8 -> 34.6 | Helped |
| Use Fish instead of GroupDRO | domain generalization models · CivilComments-Wilds test | worst-group accuracy (%): 69.1 -> 72.8 | Helped |
| Use Fish instead of Coral | domain generalization models · iWildCam-Wilds test | Macro F1: 32.8 -> 22.0 | Hurt |
| Use Fish instead of ERM | domain generalization models · Amazon-Wilds test | 10th percentile accuracy (%): 53.8 -> 53.3 | Hurt |
| Use Fish instead of ERM | domain generalization models · DomainBed average | test accuracy (%): 67.0 -> 67.1 | No clear effect |
| Group inner-loop minibatches by domain (ablation) | Fish · cDsprites (N=10) | test accuracy: 50.0 -> 100.0 | Helped |
| Group inner-loop minibatches by domain (ablation) | Fish · PACS | test accuracy: 83.9 -> 85.5 | Helped |
| Group inner-loop minibatches by domain (ablation) | Fish · OfficeHome | test accuracy: 66.5 -> 68.6 | Helped |
| Group inner-loop minibatches by domain (ablation) | Fish · VLCS | test accuracy: 77.7 -> 77.6 | No clear effect |
| Adjust the gradient-inner-product scaling parameter gamma (ablation) | SmoothFish · Wilds datasets | dataset performance | No clear effect |
| Use inner-loop minibatches from different domains rather than the same domain (ablation) | Reptile-style training · CdSprites-N | test accuracy: 50% -> 100% at N =10 | Helped |
Closest papers by meaning
- Balanced Direction from Multifarious Choices: Arithmetic Meta-Learning for Domain Generalization (CVPR 2025 · similarity 0.89)
- Gradient-Guided Annealing for Domain Generalization (CVPR 2025 · similarity 0.89)
- Lost Domain Generalization Is a Natural Consequence of Lack of Training Domains (AAAI 2024 · similarity 0.88)
- Federated Domain Generalization with Data-free On-server Matching Gradient (ICLR 2025 · similarity 0.87)
- One-Step Generalization Ratio Guided Optimization for Domain Generalization (ICML 2025 · similarity 0.87)
- Robust Domain Generalization under Divergent Marginal and Conditional Distributions (NeurIPS 2026 · similarity 0.87)
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.