Research map / Label-efficient and robust learning / Domain adaptation and generalization
Local Context-Aware Active Domain Adaptation
Tao Sun, Cheng Lu, Haibin Ling
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 diverse (clustered) selection from the top uncertain candidates (ablation) | LADA · Office-Home (5% budget) | accuracy (%): 72.6 -> 75.3 | Improved |
| add progressive anchor set augmentation with confident neighbors (ablation) | LADA · Office-Home (5% budget) | accuracy (%): 75.3 -> 77.9 | Improved |
| add RandAug mixed images to anchor set training (ablation) | LADA · Office-Home (5% budget) | accuracy (%): 79.5 -> 80.0 | Improved |
| add class balancing rejection (CBR) (ablation) | LADA · Office-Home RSUT (10% budget) | per-class average accuracy (%): 74.3 -> 75.3 | Improved |
| replace LI score with entropy, margin, confidence or NAU in LAS (ablation) | LAS · Office-Home and Office-31 | accuracy (%) | Improved |
| increase oversampling ratio M for diverse sampling (ablation) | LAS · Office-Home and Office-31 | accuracy (%) | Improved |
| choose confident samples from the neighborhood of queried data (LAA) instead of randomly (RAA) under larger domain gap (ablation) | LADA · DomainNet with added feature displacement | accuracy | Improved |
| use LAA instead of RAA on standard benchmarks (ablation) | LAS · Office-Home (10% budget) | accuracy (%) avg: 84.9 -> 84.2 | No clear effect |
| replace LAA with CDAC after LAS selection (ablation) | LADA · Office-Home (10% budget) | accuracy (%) avg: 84.2 -> 82.6 | Hurt |
| use LAS instead of other selection criteria with MME | MME · Office-Home (5% budget) | accuracy (%) avg: 77.2 -> 78.2 (vs MHPL) | Improved |
| use LADA instead of LAMDA | LADA · Office-Home (10% budget) | accuracy (%) avg: 81.9 -> 84.2 | Improved |
| use LAS with LAA instead of LAMDA on VisDA | LADA · VisDA (10% budget) | accuracy (%): 91.8 -> 93.1 | Improved |
Closest papers by meaning
- Revisiting the Domain Shift and Sample Uncertainty in Multi-source Active Domain Transfer (CVPR 2024 · similarity 0.88)
- Category-Aware Active Domain Adaptation (ICML 2024 · similarity 0.88)
- Enhancing Semi-supervised Domain Adaptation via Effective Target Labeling (AAAI 2024 · similarity 0.86)
- Gradient-Based Sample Selection for Black-Box Universal Domain Adaptation (AAAI 2025 · similarity 0.86)
- Open-Set Heterogeneous Domain Adaptation: Theoretical Analysis and Algorithm (AAAI 2025 · similarity 0.85)
- Revisiting Source-Free Domain Adaptation: Insights into Representativeness, Generalization, and Variety (CVPR 2025 · similarity 0.85)
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.
