Research map / Label-efficient and robust learning / Domain adaptation and generalization
Mixed Samples as Probes for Unsupervised Model Selection in Domain Adaptation
Dapeng Hu, Jian Liang, Jun Hao Liew, Chuhui Xue, Song Bai, Xinchao Wang
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 |
|---|---|---|---|
| use MixVal instead of SourceVal for model selection | UDA model selection · VisDA | validation accuracy (%): 71.04 -> 77.86 | Improved |
| use MixVal instead of SND for model selection | UDA model selection · Office-Home | validation accuracy (%): 66.02 -> 69.84 | Improved |
| use MixVal instead of SND for model selection | BNM · DomainNet | validation accuracy (%): 56.93 -> 67.85 | Improved |
| use MixVal instead of Entropy for model selection | DANCE open-partial-set · Office-Home | HOS (%): 34.27 -> 67.01 | Improved |
| use MixVal instead of Entropy for model selection | SHOT source-free · VisDA | accuracy (%): 82.65 -> 83.12 | Improved |
| use MixVal instead of SND for model selection with two hyperparameters | MDD · Office-Home | validation accuracy (%): 57.42 -> 70.25 | Improved |
| use MixVal instead of SND for model selection with two hyperparameters | MCC · Office-Home | validation accuracy (%): 69.96 -> 73.75 | Improved |
| use MixVal instead of SND for validating mixup-based UDA | DMRL · Office-31 | validation accuracy (%): 77.55 -> 78.16 | Improved |
| use MixVal instead of SND on a segmentation task | AdaptSeg · GTAV to Cityscapes | mIoU (%): 40.69 -> 42.20 | Improved |
| use MixVal instead of Entropy with ViT-B backbone | BNM · DomainNet R to S | validation accuracy (%): 28.21 -> 54.78 | Improved |
| use Entropy instead of SourceVal for segmentation model selection | AdvEnt · GTAV to Cityscapes | mIoU (%): 39.08 -> 38.41 | Hurt |
| use RV instead of SourceVal for partial-set model selection | PADA · Office-Home partial-set | validation accuracy (%): 58.87 -> 58.34 | Hurt |
Closest papers by meaning
- Towards Reliable Model Selection for Unsupervised Domain Adaptation: An Empirical Study and A Certified Baseline (NeurIPS 2024 · similarity 0.91)
- Towards Accurate Model Selection in Deep Unsupervised Domain Adaptation (ICML 2019 · similarity 0.89)
- Can We Evaluate Domain Adaptation Models Without Target-Domain Labels? (ICLR 2024 · similarity 0.88)
- Pseudo-Calibration: Improving Predictive Uncertainty Estimation in Unsupervised Domain Adaptation (ICML 2024 · similarity 0.87)
- Gradient-Based Sample Selection for Black-Box Universal Domain Adaptation (AAAI 2025 · similarity 0.86)
- Pseudo-Labeling for Unsupervised Domain Adaptation with Kernel GLMs (NeurIPS 2026 · 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.
