Research map / Label-efficient and robust learning / Domain adaptation and generalization
Exploiting Inter-Sample Affinity for Knowability-Aware Universal Domain Adaptation
Yifan Wang, Lin Zhang, Ran Song, Hongliang Li, Paul L. Rosin, Wei 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 source-domain loss L_s (ablation) | full method · Office-31 UniDA | H-score: 36.4 -> 88.7 | Improved |
| add unknown-affinity loss L_unk (ablation) | full method · Office-31 UniDA | H-score: 82.6 -> 88.7 | Improved |
| add uncertainty loss L_unc (ablation) | full method · Office-31 UniDA | H-score: 86.3 -> 88.7 | Improved |
| add known-affinity loss L_k (ablation) | full method · Office-31 UniDA | H-score: 86.5 -> 88.7 | Improved |
| increase neighbors k from 10 to 90 (ablation) | full method · Office-31 W2D | H-score: 98.3 -> 92.8 | Hurt |
| increase neighbors k from 10 to 90 (ablation) | full method · Office-31 A2D | H-score: 87.4 -> 84.0 | Hurt |
| scale threshold c_tau from 0.1 to 0.8 (ablation) | full method · Office-31 UniDA | H-score: 87.9 -> 88.4 | No clear effect |
| vary lambda weight on target losses (ablation) | full method · Office-31 UniDA | H-score | No clear effect |
| swap method to ours vs OVANet on new Art samples | Ours vs OVANet · Office-31 to OfficeHome Art | H-score: 88.2 -> 96.9 | Improved |
| swap method to ours vs OVANet on VGGNet | Ours vs OVANet · Office-31 ODA (VGGNet) | H-score: 88.7 -> 90.6 | Improved |
| use proposed method vs OVANet on OfficeHome | Ours vs OVANet · OfficeHome UniDA | H-score: 71.8 -> 74.6 | Improved |
Closest papers by meaning
- Gradient-Based Sample Selection for Black-Box Universal Domain Adaptation (AAAI 2025 · similarity 0.91)
- Distinguish Then Exploit: Source-free Open Set Domain Adaptation via Weight Barcode Estimation and Sparse Label Assignment (CVPR 2025 · similarity 0.89)
- Universal Semi-Supervised Domain Adaptation by Mitigating Common-Class Bias (CVPR 2024 · similarity 0.88)
- LEAD: Learning Decomposition for Source-free Universal Domain Adaptation (CVPR 2024 · similarity 0.88)
- Category-Aware Active Domain Adaptation (ICML 2024 · similarity 0.88)
- Tackling Dimensional Collapse toward Comprehensive Universal Domain Adaptation (ICML 2025 · 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.
