Research map / Label-efficient and robust learning / Domain adaptation and generalization
Failure Modes of Domain Generalization Algorithms
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 |
|---|---|---|---|
| Freeze BYOL representations rather than train representations from random initialization | ERM on Camelyon17 · Camelyon17 test domains | error: 0.287 -> 0.077 | Helped |
| Set weight decay to 0 | ERM with Frozen BYOL on Camelyon17 · Camelyon17 test domains | error: 0.077 -> 0.070 | Helped |
| Add Spectral Decoupling with β=5 | ERM with Frozen BYOL on Camelyon17 · Camelyon17 test domains | error: 0.077 -> 0.070 | Helped |
| Use GroupDRO with β=5 instead of ERM | Frozen BYOL representations on Camelyon17 · Camelyon17 test domains | error: 0.077 -> 0.078 | Hurt |
| Use IRM with β=5 instead of ERM | Frozen BYOL representations on Camelyon17 · Camelyon17 test domains | error: 0.077 -> 0.081 | Hurt |
| Increase ERM+HSIC regularization strength toward β=1 | ERM+HSIC on Colored MNIST · Colored MNIST test domains | error: 0.09 | Helped |
| Increase ERM+HSIC regularization beyond the point where representations collapse | ERM+HSIC on Colored MNIST · Colored MNIST training domains | error | Hurt |
| Increase domain invariance during training | Domain generalization algorithms on Camelyon17 · Camelyon17 validation domains | generalization error | No clear effect |
| Use a larger ResNet model | ERM+HSIC on Colored MNIST (β=1) · Colored MNIST validation domains | training-test domain distinguishability | Hurt |
| Select models using validation-domain performance | Domain generalization models on Colored MNIST · Colored MNIST validation and test domains | validation-test error correlation: 0.98, 0.8, 0.73 and 0.31 | Helped |
Closest papers by meaning
- Lost Domain Generalization Is a Natural Consequence of Lack of Training Domains (AAAI 2024 · similarity 0.92)
- Sharpness-Aware Model-Agnostic Long-Tailed Domain Generalization (AAAI 2024 · similarity 0.89)
- Robust Domain Generalization under Divergent Marginal and Conditional Distributions (NeurIPS 2026 · similarity 0.89)
- Gradient-Guided Annealing for Domain Generalization (CVPR 2025 · similarity 0.89)
- Unknown Domain Inconsistency Minimization for Domain Generalization (ICLR 2024 · similarity 0.89)
- Mixup-Induced Domain Extrapolation for Domain Generalization (AAAI 2024 · 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.