Research map / Label-efficient and robust learning / Domain adaptation and generalization
Adaptive Methods for Real-World Domain Generalization
Check 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 |
|---|---|---|---|
| Append a domain prototype to image features | DA-ERM · LT-ImageNet test domains | top-1 accuracy: 53.7% vs. 56.1% | Helped |
| Append a domain prototype to image features | DA-ERM · Geo-YFCC test domains | top-1 accuracy: 22.4 -> 23.4 | Helped |
| Append a domain prototype to image features | DA-CORAL · Average across five DomainBed datasets | accuracy: 63.8 -> 64.7 | Helped |
| Add domain-adaptive classification to MMD | MMD · Average across five DomainBed datasets | accuracy: 62.2% vs. 64.5% | Helped |
| Add domain-adaptive classification to CORAL | CORAL · Average across five DomainBed datasets | accuracy: +0.9% | Helped |
| Use an incorrect random domain embedding instead of the correct embedding (ablation) | DA-ERM · LT-ImageNet test domains | top-1 accuracy: 80.5 -> 75.3 | Hurt |
| Use an incorrect random domain embedding instead of the correct embedding (ablation) | DA-ERM · Geo-YFCC test domains | top-1 accuracy: 23.4 -> 21.3 | Hurt |
| Replace prototypical embeddings with average ERM pool5 features (ablation) | DA-ERM · LT-ImageNet test domains | top-1 accuracy: 80.5 -> 77.9 | Hurt |
| Replace prototypical embeddings with domain-ID softmax features (ablation) | DA-ERM · LT-ImageNet test domains | top-1 accuracy: 80.5 -> 78.8 | Hurt |
| Replace prototypical embeddings with domain-ID softmax features (ablation) | DA-ERM · Geo-YFCC test domains | top-1 accuracy: 23.4 -> 23.5 | Helped |
| Vary the number of prototype construction points from 50 up to 2K (ablation) | DA-ERM · Geo-YFCC validation set | top-1 accuracy: 23.7/49.0; 23.3/49.1; 23.4/49.3; 23.6/49.3; 23.4/49.1; 23.4/49.2 | No clear effect |
| Vary the number of domains sampled per prototypical-training round across [3, 8] (ablation) | Prototypical domain embedding · Domain generalization benchmarks | accuracy | No clear effect |
Closest papers by meaning
- Lost Domain Generalization Is a Natural Consequence of Lack of Training Domains (AAAI 2024 · similarity 0.90)
- Set Valued Predictions For Robust Domain Generalization (ICML 2025 · similarity 0.89)
- Enhancing Evolving Domain Generalization through Dynamic Latent Representations (AAAI 2024 · similarity 0.89)
- When Domain Generalization meets Generalized Category Discovery: An Adaptive Task-Arithmetic Driven Approach (CVPR 2025 · similarity 0.89)
- A General Representation-Based Approach to Multi-Source Domain Adaptation (ICML 2025 · similarity 0.89)
- Robust Domain Generalization under Divergent Marginal and Conditional Distributions (NeurIPS 2026 · similarity 0.89)
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.