Research map / Label-efficient and robust learning / Domain adaptation and generalization
PAS: Estimating the target accuracy before domain adaptation
Raphaella Diniz, Jackson de Faria, Martin Ester
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 PAS instead of MMD to predict target accuracy | source selection for UDA · All benchmarks | Pearson correlation: 0.37 -> 0.83 | Improved |
| use PAS instead of A-distance | source selection for UDA · All benchmarks | Pearson correlation: 0.04 -> 0.83 | Improved |
| use cosine distance to class centroid instead of Euclidean distance (ablation) | PAS · All benchmarks | Pearson correlation: 0.68 -> 0.79 | Improved |
| use cosine distance to class centroid instead of Euclidean distance (ablation) | PAS · Office-31 | Pearson correlation: 0.69 vs 0.63 | Hurt |
| use class centroid instead of average cosine distance to source samples (ablation) | PAS · All benchmarks | Pearson correlation: 0.66 -> 0.79 | Improved |
| apply PAS to multi-object images | PAS · ImageCLEF | Pearson correlation: 0.44 overall; 0.22 for ResNet-50 | Hurt |
| use subset of samples to compute PAS (ablation) | PAS · Office-Home | PAS value / ranking | No clear effect |
| use PAS to rank pre-trained backbones with fixed DA method | PAS with DANN and MCC · Office-Home A->C, Office-31 W->A | target accuracy ranking | Improved |
| use PAS instead of MMD on DomainNet | source selection for UDA · DomainNet | Pearson correlation: -0.09 -> 0.53 | Improved |
Closest papers by meaning
- Rethinking the Role of Pre-Trained Networks in Source-Free Domain Adaptation (ICCV 2023 · similarity 0.90)
- Source-Free Domain Adaptation Guided by Vision and Vision-Language Pre-training (IJCV 2024 · similarity 0.90)
- Can We Evaluate Domain Adaptation Models Without Target-Domain Labels? (ICLR 2024 · similarity 0.87)
- Invertible Projection and Conditional Alignment for Multi-Source Blended-Target Domain Adaptation (AAAI 2025 · similarity 0.87)
- Understanding and Improving Source-free Domain Adaptation from a Theoretical Perspective (CVPR 2024 · similarity 0.87)
- Revisiting Source-Free Domain Adaptation: Insights into Representativeness, Generalization, and Variety (CVPR 2025 · similarity 0.87)
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.
