Research map / Label-efficient and robust learning / Domain adaptation and generalization
Towards Unsupervised Domain Adaptation via Domain-Transformer
Chuan-Xian Ren, Yiming Zhai, You-Wei Luo, Hong Yan
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 Ls, Lt and Lent local-structure losses to domain-level attention (ablation) | DoT · VisDA-2017 S->R | accuracy (%): 81.4 -> 84.9 | Improved |
| add Ls, Lt and Lent local-structure losses (ablation) | DoT · ImageCLEF C->I | accuracy (%): 92.8 -> 95.3 | Improved |
| add Ls, Lt and Lent local-structure losses (ablation) | DoT · ImageCLEF I->C | accuracy (%): 96.0 -> 97.8 | Improved |
| add Ls, Lt and Lent local-structure losses (ablation) | DoT · Office-31 A->W | accuracy (%): 93.6 -> 94.8 | Improved |
| add Ls, Lt and Lent local-structure losses (ablation) | DoT · Office-31 W->A | accuracy (%): 82.2 -> 83.1 | Improved |
| add supervised locality term Ls alone (ablation) | DoT · VisDA-2017 S->R | accuracy (%): 81.4 -> 82.5 | Improved |
| add Ls alone (ablation) | DoT · Office-31 A->W | accuracy (%): 93.6 -> 93.5 | No clear effect |
| add Lt and Lent without Ls (ablation) | DoT · VisDA-2017 S->R | accuracy (%): 81.4 -> 83.3 | Improved |
| add Ls on top of Lt and Lent (ablation) | DoT · VisDA-2017 S->R | accuracy (%): 83.3 -> 84.9 | Improved |
| use domain-level attention instead of patch-level attention (ablation) | DoT · VisDA-2017 S->R | accuracy (%): 68.8 -> 84.9 | Improved |
| use domain-level attention instead of patch-level attention (ablation) | DoT · Office-31 A->W | accuracy (%): 85.5 -> 94.8 | Improved |
| use domain-level attention instead of patch-level attention (ablation) | DoT · Office-31 W->A | accuracy (%): 69.4 -> 83.1 | Improved |
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.
- DeepJDOT — DeepJDOT: Deep Joint Distribution Optimal Transport for Unsupervised Domain Adaptation (ECCV 2018) · better on 1 of 1 shared result
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
- CDTrans: Cross-domain Transformer for Unsupervised Domain Adaptation (ICLR 2022 · similarity 0.90)
- Unsupervised Domain Adaptation by Matching Distributions Based on the Maximum Mean Discrepancy via Unilateral Transformations (AAAI 2019 · similarity 0.89)
- Structure-Aware Feature Fusion for Unsupervised Domain Adaptation (AAAI 2020 · similarity 0.89)
- Inferring Latent Domains for Unsupervised Deep Domain Adaptation (TPAMI 2019 · similarity 0.89)
- Enhanced Transport Distance for Unsupervised Domain Adaptation (CVPR 2020 · similarity 0.89)
- Can We Evaluate Domain Adaptation Models Without Target-Domain Labels? (ICLR 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.
