Research map / Label-efficient and robust learning / Domain adaptation and generalization
Domain Generalization by Mutual-Information Regularization with Pre-trained Models
Junbum Cha, Kyungjae Lee, Sungrae Park, Sanghyuk Chun
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add MIRO mutual-information regularization to ERM | ERM ResNet-50 ImageNet · DomainBed average | out-of-domain accuracy (%): 64.2 -> 65.9 | Improved |
| add MIRO to ERM | ERM RegNetY-16GF SWAG · DomainBed average | out-of-domain accuracy (%): 68.0 -> 74.1 | Improved |
| add MIRO to ERM | ERM ViT-B CLIP · DomainBed average | out-of-domain accuracy (%): 61.1 -> 73.7 | Improved |
| add MIRO to ERM | ERM ResNet-50 CLIP · DomainBed average | out-of-domain accuracy (%): 44.9 -> 61.2 | Improved |
| add MIRO to ERM | ERM Barlow Twins · DomainBed average | out-of-domain accuracy (%): 58.4 -> 61.9 | Improved |
| add MIRO to SWAD | ERM + SWAD · DomainBed average | out-of-domain accuracy (%): 66.9 -> 68.1 | Improved |
| add MIRO to SWAD with RegNet | ERM + SWAD RegNetY-16GF · DomainBed average | out-of-domain accuracy (%): 73.2 -> 77.3 | Improved |
| add MIRO to ERM | ERM MoCo v3 · PACS | out-of-domain accuracy (%): 86.7 -> 86.3 | Hurt |
| naively fine-tune CLIP ViT-B instead of ImageNet ResNet-50 | ERM · DomainBed average | out-of-domain accuracy (%): 64.2 -> 61.1 | Hurt |
| add MIRO to ERM | ERM ResNet-50 · Camelyon17 average | accuracy (%): 94.9 -> 95.6 | Improved |
| add pre-trained weights (ERM+ vs ERM-) (ablation) | ERM ResNet-50 · PACS | accuracy (%): 51.6 -> 84.2 | Improved |
| use class-conditional MI (C-MIRO) instead of MIRO (ablation) | MIRO · DomainBed 4 datasets | accuracy (%): 71.3 -> 71.0 | No clear effect |
Closest papers by meaning
- Learning To Diversify for Single Domain Generalization (ICCV 2021 · similarity 0.90)
- Model-Based Domain Generalization (NeurIPS 2021 · similarity 0.89)
- Robust Domain Generalization under Divergent Marginal and Conditional Distributions (NeurIPS 2026 · similarity 0.89)
- Unknown Domain Inconsistency Minimization for Domain Generalization (ICLR 2024 · similarity 0.88)
- Enhancing Evolving Domain Generalization through Dynamic Latent Representations (AAAI 2024 · similarity 0.88)
- Improving Domain Generalization with Domain Relations (ICLR 2024 · similarity 0.88)
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.
