Research map / Label-efficient and robust learning / Domain adaptation and generalization
Adversarial Unsupervised Domain Adaptation With Conditional and Label Shift: Infer, Align and Iterate
Xiaofeng Liu, Zhenhua Guo, Site Li, Fangxu Xing, Jane You, C.-C. Jay Kuo, Georges El Fakhri, Jonghye Woo
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 joint parameterization of discriminator and classifier (ablation) | CLS · VisDA17-val | mean accuracy: 80.8 -> 81.6 | Improved |
| add fade-in of estimated target label distribution p_t(y) (ablation) | CLS · VisDA17-val | mean accuracy: 81.3 -> 81.6 | Improved |
| add joint parameterization to CLS+TDDA (ablation) | CLS+TDDA · VisDA17-val | mean accuracy: 82.2 -> 82.9 | Improved |
| combine CLS with TDDA | CLS · VisDA17-val | mean accuracy: 81.6 -> 82.9 | Improved |
| use ResNet152 backbone instead of ResNet101 | CLS · VisDA17-val | mean accuracy: 81.6 -> 83.8 | Improved |
| add joint parameterization (adversarial base) (ablation) | CLS · GTA5 to Cityscapes | mIoU: 50.8 -> 51.2 | Improved |
| add joint parameterization (self-training combination) (ablation) | CLS · GTA5 to Cityscapes | mIoU: 52.5 -> 53.0 | Improved |
| add joint parameterization on Office-31 partial UDA (ablation) | CLS · Office-31 partial UDA (ResNet-50) | mean accuracy: 97.94 -> 98.24 | Improved |
| add joint parameterization on ImageNet-Caltech partial UDA (ablation) | CLS · ImageNet-Caltech partial UDA (ResNet-50) | average accuracy: 83.52 -> 84.76 | Improved |
| replace DANN with CLS (conditional and label alignment) | DANN · VisDA17-val | mean accuracy: 57.4 -> 81.6 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- PANDA — PANDA: Adapting Pretrained Features for Anomaly Detection and Segmentation (CVPR 2021) · better on 1 of 1 shared result
- DCAN — DCAN: Dual Channel-wise Alignment Networks for Unsupervised Scene 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
- CASUAL: Conditional Support Alignment for Domain Adaptation with Label Shift (AAAI 2025 · similarity 0.95)
- Unsupervised Domain Adaptation via Regularized Conditional Alignment (ICCV 2019 · similarity 0.92)
- Can We Evaluate Domain Adaptation Models Without Target-Domain Labels? (ICLR 2024 · similarity 0.92)
- Unsupervised Robust Domain Adaptation: Paradigm, Theory and Algorithm (IJCV 2025 · similarity 0.92)
- On Unsupervised Domain Adaptation: Pseudo Label Guided Mixup for Adversarial Prompt Tuning (AAAI 2024 · similarity 0.90)
- Toward Improving Robustness and Accuracy in Unsupervised Domain Adaptation (AAAI 2025 · similarity 0.90)
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.
