Research map / Label-efficient and robust learning / Domain adaptation and generalization
DomainDrop: Suppressing Domain-Sensitive Channels for Domain Generalization
Jintao Guo, Lei Qi, Yinghuan Shi
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 |
|---|---|---|---|
| apply DomainDrop to all layers (ablation) | ResNet-18 baseline · PACS | average accuracy (%): 81.00 -> 85.06 | Improved |
| add DomainDrop to OfficeHome ResNet-18 baseline (ablation) | ResNet-18 baseline · OfficeHome | average accuracy (%): 63.59 -> 65.30 | Improved |
| add layer-wise training scheme (ablation) | DomainDrop · PACS | average accuracy (%): 85.06 -> 85.94 | Improved |
| add dual consistency loss (ablation) | DomainDrop · PACS | average accuracy (%): 85.06 -> 85.74 | Improved |
| combine layer-wise training and consistency loss with DomainDrop (ablation) | DomainDrop · PACS | average accuracy (%): 85.94 -> 86.66 | Improved |
| set dropout ratio to 0.33 (ablation) | DomainDrop · PACS | average accuracy (%) | No clear effect |
| apply DomainDrop with probability at least 0.7 (ablation) | DomainDrop · PACS | average accuracy (%) | No clear effect |
| add DomainDrop to FACT | FACT · PACS (ResNet-18) | average accuracy (%): 84.61 -> 87.24 | Improved |
| add DomainDrop to RandAug | RandAug · PACS (ResNet-18) | average accuracy (%): 83.63 -> 86.69 | Improved |
| add DomainDrop to GFNet | GFNet · PACS | average accuracy (%): 87.76 -> 91.02 | Improved |
| apply DomainDrop to reduce domain divergence | ResNet-18 · PACS | source-target divergence (x10): 7.25 -> 4.09 | Improved |
| use traditional dropout (Cutout, C-Drop) on source domains | ResNet-18 · PACS | average accuracy (%): 81.00 -> 80.60 (Cutout), 81.11 (C-Drop) | No clear effect |
Later papers that compared with this method
1 later accepted paper reports DomainDrop as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- Representation Enhancement-Stabilization: Reducing Bias-Variance of Domain Generalization (ECCV 2024 · better on 1 of 2 shared results)
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- StyleNeophile — Style Neophile: Constantly Seeking Novel Styles for Domain Generalization (CVPR 2022) · better on 1 of 1 shared result
- SWAD — SWAD: Domain Generalization by Seeking Flat Minima (NeurIPS 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- ConstStyle: Robust Domain Generalization with Unified Style Transformation (ICCV 2025 · similarity 0.89)
- Domain Generalization by Learning and Removing Domain-specific Features (NeurIPS 2022 · similarity 0.89)
- Deep Domain Generalization via Conditional Invariant Adversarial Networks (ECCV 2018 · similarity 0.88)
- Robust Domain Generalization under Divergent Marginal and Conditional Distributions (NeurIPS 2026 · similarity 0.87)
- Set Valued Predictions For Robust Domain Generalization (ICML 2025 · similarity 0.86)
- Sharpness-Aware Model-Agnostic Long-Tailed Domain Generalization (AAAI 2024 · similarity 0.86)
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.
