Research map / Label-efficient and robust learning / Domain adaptation and generalization
Sharpness-Aware 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 SAGM instead of the strongest listed baseline, SagNet | Domain generalization on PACS · PACS | accuracy: 86.3% → 86.6 | Helped |
| Use SAGM instead of mDSDI | Domain generalization on VLCS · VLCS | accuracy: 79.0% → 80.0% | Helped |
| Use SAGM instead of Miro (with CLIP) | Domain generalization on OfficeHome · OfficeHome | accuracy: 70.5 → 70.1 | Hurt |
| Use SAGM instead of Miro (with CLIP) | Domain generalization on TerraInc · TerraInc | accuracy: 50.4 → 48.8 | Hurt |
| Use SAGM instead of Miro (with CLIP) | Domain generalization on DomainNet · DomainNet | accuracy: 44.3 → 45.0 | Helped |
| Add gradient matching to ERM + SAM objective (ablation) | SAGM · DomainBed average | accuracy: 64.8 → 66.1 | Helped |
| Use SAM's perturbed loss only instead of SAGM (ablation) | SAGM · DomainBed average | accuracy: 66.1 vs 64.5 | Hurt |
| Use GSAM's perturbed-loss and surrogate-gap objective instead of SAGM (ablation) | SAGM · DomainBed average | accuracy: 66.1 vs 65.1 | Hurt |
| Use ERM only instead of SAGM (ablation) | SAGM · DomainBed average | accuracy: 66.1 vs 63.9 | Hurt |
| Add Mixstyle augmentation | SAGM on PACS · PACS out-of-domain | accuracy: 86.2 → 87.2 | Helped |
| Add Mixup augmentation | ERM on PACS · PACS out-of-domain | accuracy: 85.3 → 84.8 | Hurt |
| Add CutMix augmentation | ERM on PACS · PACS out-of-domain | accuracy: 85.3 → 83.8 | Hurt |
Closest papers by meaning
- Unknown Domain Inconsistency Minimization for Domain Generalization (ICLR 2024 · similarity 0.91)
- Seeking Consistent Flat Minima for Better Domain Generalization via Refining Loss Landscapes (CVPR 2025 · similarity 0.90)
- Gradient-Guided Annealing for Domain Generalization (CVPR 2025 · similarity 0.90)
- Domain-Inspired Sharpness-Aware Minimization Under Domain Shifts (ICLR 2024 · similarity 0.89)
- One-Step Generalization Ratio Guided Optimization for Domain Generalization (ICML 2025 · similarity 0.89)
- Friendly Sharpness-Aware Minimization (CVPR 2024 · 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.