Research map / Label-efficient and robust learning / Domain adaptation and generalization
Incorporating Pre-Training Data Matters in Unsupervised Domain Adaptation
Yinsong Xu, Aidong Men, Yang Liu, Xiahai Zhuang, Qingchao Chen
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 pre-training data cross-entropy loss Lp (ablation) | SHOT · Office-Home Cl->Pr | accuracy: 78.2 -> 79.9 | Improved |
| add Lp (ablation) | SHOT · Office-Home Cl->Ar | accuracy: 68.0 -> 69.4 | Improved |
| add intermediate-domain semantic loss Lsem alone (ablation) | SHOT · Office-Home Cl->Pr | accuracy: 78.2 -> 79.9 | Improved |
| add Lsem on top of Lp (ablation) | SHOT · Office-Home Cl->Pr | accuracy: 79.9 -> 81.2 | Improved |
| add feature interpolation loss Lfeat (ablation) | SHOT with Lp and Lsem · Office-Home Cl->Pr | accuracy: 81.2 -> 81.3 | Improved |
| select WordNet-similar classes instead of all 1000 classes (ablation) | TriDA+SHOT · Office-Home | accuracy: 72.3 -> 73.5 | Improved |
| select classes by similarity instead of random (ablation) | TriDA+SHOT · Office-Home | accuracy: 72.6 -> 73.5 | Improved |
| vary Beta(alpha,alpha) mixing parameter from 0.5 to 4.0 (ablation) | TriDA+SHOT · Office-Home | avg accuracy: 73.2 -> 73.5 | No clear effect |
| add TriDA to CDAN | CDAN · Office-Home (ResNet-50, UDA) | accuracy: 68.8 -> 70.9 | Improved |
| add TriDA to MDD | MDD · Office-Home (ResNet-50, UDA) | accuracy: 69.7 -> 71.4 | Improved |
| add TriDA to SHOT | SHOT · Office-Home (ResNet-50, SFUDA) | accuracy: 71.8 -> 73.5 | Improved |
| add TriDA to CoWA | CoWA · Office-31 (ResNet-50) | accuracy: 90.3 -> 90.3 | No clear effect |
Closest papers by meaning
- Can We Evaluate Domain Adaptation Models Without Target-Domain Labels? (ICLR 2024 · similarity 0.91)
- Prior Knowledge Guided Unsupervised Domain Adaptation (ECCV 2022 · similarity 0.91)
- Moment Matching for Multi-Source Domain Adaptation (ICCV 2019 · similarity 0.91)
- UDA-Bench: Revisiting Common Assumptions in Unsupervised Domain Adaptation Using a Standardized Framework (ECCV 2024 · similarity 0.91)
- Enhancing Domain Adaptation through Prompt Gradient Alignment (NeurIPS 2024 · similarity 0.89)
- Toward Improving Robustness and Accuracy in Unsupervised Domain Adaptation (AAAI 2025 · similarity 0.89)
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.
