Research map / Label-efficient and robust learning / Domain adaptation and generalization
Label Propagation with Augmented Anchors: A Simple Semi-Supervised Learning baseline for Unsupervised Domain Adaptation
Yabin Zhang, Bin Deng, Kui Jia, Lei Zhang
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 augmented anchors (virtual instances) to label propagation | LP · Office-31 (average of 6 tasks) | accuracy: 84.0 -> 87.4 | Improved |
| add augmented anchors (virtual instances) to label propagation | LP · ImageCLEF-DA (average of 6 tasks) | accuracy: 84.6 -> 88.9 | Improved |
| add augmented anchors (virtual instances) to label propagation | LP (ResNet-50) · VisDA-2017 (synthetic to real) | accuracy: 69.8 -> 78.7 | Improved |
| replace pseudo label generator with A2LP | MSTN · VisDA-2017 (synthetic to real) (ResNet-50) | accuracy: 71.9 -> 81.5 | Improved |
| replace pseudo label generator with A2LP | MSTN · Office-31 (average of 6 tasks) | accuracy: 88.5 -> 89.8 | Improved |
| replace pseudo label generator with A2LP | CAN · Office-31 (average of 6 tasks) | accuracy: 90.3 -> 90.7 | Improved |
| replace pseudo label generator with A2LP | CAN · VisDA-2017 (synthetic to real) (ResNet-101) | accuracy: 87.2 -> 87.6 | No clear effect |
| add entropy-based instance weights (ablation) | A2LP · Office-31 A to W | accuracy: 87.4 -> 87.7 | Improved |
| add entropy-based instance weights (ablation) | A2LP · Office-31 W to A | accuracy: 75.4 -> 75.9 | Improved |
| use all source instances instead of K source class centers (ablation) | A2LP · VisDA-2017-Small | accuracy: 77.9 -> 79.3 | Improved |
| increase noise level of initial label predictions from 0% to 100% (ablation) | A2LP · ImageCLEF-DA P to C | accuracy: 92.8 -> 90.0 | Hurt |
| raise alpha from 0.75 to 2.0 (ablation) | A2LP · ImageCLEF-DA P to C | accuracy: 95.8 -> 16.8 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DADA — DADA: Depth-Aware Domain Adaptation in Semantic Segmentation (ICCV 2019) · better on 2 of 2 shared results
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
- Universal Domain Adaptation (CVPR 2019 · similarity 0.91)
- Universal Semi-Supervised Domain Adaptation by Mitigating Common-Class Bias (CVPR 2024 · similarity 0.89)
- Improving Unsupervised Domain Adaptation: A Pseudo-Candidate Set Approach (ECCV 2024 · similarity 0.89)
- Can We Evaluate Domain Adaptation Models Without Target-Domain Labels? (ICLR 2024 · similarity 0.88)
- A Unified Framework for Heterogeneous Semi-supervised Learning (CVPR 2025 · similarity 0.88)
- Enhancing Semi-supervised Domain Adaptation via Effective Target Labeling (AAAI 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.
