Research map / Label-efficient and robust learning / Domain adaptation and generalization
GeT: Generative Target Structure Debiasing for Domain Adaptation
Can Zhang, Gim Hee Lee
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 |
|---|---|---|---|
| replace nearest-centroid classifier with online generative (GMM) classifier for pseudo labels (ablation) | GeT · VisDA-2017 | average accuracy (%): 80.3 -> 81.4 | Improved |
| add embedding-level structure similarity regularization (L_KL^f) (ablation) | GeT · VisDA-2017 | average accuracy (%): 80.6 -> 83.4 | Improved |
| add mixed soft labels from generative classifier and auxiliary distribution (ablation) | GeT · VisDA-2017 | average accuracy (%): 82.0 -> 83.4 | Improved |
| add structure similarity regularization and mixing on Office-31 (ablation) | GeT · Office-31 | average accuracy (%): 84.0 -> 86.0 | Improved |
| add GeT to source-only model | ResNet-50 · Office-Home UDA | mean accuracy (%): 70.2 -> 73.5 (vs ATDOC-NC) | Improved |
| add GeT to CDAN+E | CDAN+E · Office-Home UDA | mean accuracy (%): 68.5 -> 74.0 | Improved |
| use GeT as pseudo-label module for MixMatch | MixMatch · Office-31 | average accuracy (%): 82.9 -> 88.65 | Improved |
| add class distribution bias to target domain | all methods incl. ATDOC · Office-Home UDA imbalanced | average accuracy (%): ATDOC-NC 70.2 -> 67.1; GeT 70.2 on imbalanced | Hurt |
| GeT versus ATDOC-NC under imbalanced PDA | GeT · Office-Home PDA imbalanced | average accuracy (%): 70.9 -> 75.3 | Improved |
| vary mixup coefficient gamma_Q and memory decay gamma_pi (ablation) | GeT · A->W imbalanced Office-31 | average accuracy (%) | No clear effect |
| GeT versus S3D on DomainNet-126 semi-supervised DA | GeT · DomainNet-126 SSDA | average accuracy (%): 69.9 -> 72.2 (1-shot) | 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.
- FixBi — FixBi: Bridging Domain Spaces for Unsupervised Domain Adaptation (CVPR 2021) · better on 1 of 1 shared result
- MixMatch — MixMatch: A Holistic Approach to Semi-Supervised Learning (NeurIPS 2019) · 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
- De-Confusing Pseudo-Labels in Source-Free Domain Adaptation (ECCV 2024 · similarity 0.89)
- Universal Source-Free Domain Adaptation (CVPR 2020 · similarity 0.89)
- Enhancing Semi-supervised Domain Adaptation via Effective Target Labeling (AAAI 2024 · similarity 0.88)
- Mapping conditional distributions for domain adaptation under generalized target shift (ICLR 2022 · similarity 0.88)
- Diffusion-Driven Progressive Target Manipulation for Source-Free Domain Adaptation (NeurIPS 2025 · similarity 0.88)
- 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.
