Research map / Label-efficient and robust learning / Domain adaptation and generalization
Deep Co-Training With Task Decomposition for Semi-Supervised Domain Adaptation
Luyu Yang, Yan Wang, Mingfei Gao, Abhinav Shrivastava, Kilian Q. Weinberger, Wei-Lun Chao, Ser-Nam Lim
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 task decomposition with co-training (DeCoTa instead of MiST) (ablation) | MiST · DomainNet (3-shot) | accuracy: 74.6 -> 75.6 | Improved |
| add task decomposition with co-training (DeCoTa instead of MiST) (ablation) | MiST · Office-Home (3-shot, VGG-16) | accuracy: 63.9 -> 69.3 | Improved |
| add MixUp to pseudo-label self-training (ablation) | MiST · DomainNet (3-shot) | accuracy: 65.6 -> 74.6 | Improved |
| add co-training exchange of pseudo-labels to two-view task decomposition (ablation) | Two-view MiST · DomainNet (3-shot) | accuracy: 72.0 -> 75.6 | Improved |
| add co-training over a vanilla ensemble of two MiST models (ablation) | Vanilla-Ensemble · DomainNet (3-shot) | accuracy: 74.7 -> 75.6 | Improved |
| add co-training over a vanilla ensemble of two MiST models (ablation) | Vanilla-Ensemble · Office-Home (3-shot, VGG-16) | accuracy: 64.7 -> 69.3 | Improved |
| add two-way teaching instead of w_f only (ablation) | DeCoTa · DomainNet (3-shot) | accuracy: 69.1 -> 75.6 | Improved |
| add two-way teaching instead of w_g only (ablation) | DeCoTa · DomainNet (3-shot) | accuracy: 72.9 -> 75.6 | Improved |
| add co-training (ablation) | DeCoTa w_f · DomainNet (3-shot) | accuracy: 67.0 -> 75.2 | Improved |
| replace best prior method with DeCoTa | UODA · DomainNet (1-shot) | accuracy: 68.5 -> 73.9 | Improved |
| replace UODA with DeCoTa | UODA · Office-Home (1-shot, VGG-16) | accuracy: 63.9 -> 63.3 | Hurt |
| replace FixMatch with DeCoTa | FixMatch · DomainNet (50-shot) | accuracy: 79.2 -> 80.8 | 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.
- FixMatch — FixMatch: Simplifying Semi-Supervised Learning with Consistency and Confidence (NeurIPS 2020) · 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
- AdaMatch: A Unified Approach to Semi-Supervised Learning and Domain Adaptation (ICLR 2022 · similarity 0.91)
- Semi-Supervised Domain Adaptation With Source Label Adaptation (CVPR 2023 · similarity 0.89)
- Forget More to Learn More: Domain-specific Feature Unlearning for Semi-supervised and Unsupervised Domain Adaptation (ECCV 2024 · similarity 0.89)
- Universal Semi-Supervised Domain Adaptation by Mitigating Common-Class Bias (CVPR 2024 · similarity 0.88)
- Enhancing Semi-supervised Domain Adaptation via Effective Target Labeling (AAAI 2024 · similarity 0.88)
- A Unified Framework for Heterogeneous Semi-supervised Learning (CVPR 2025 · similarity 0.86)
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.
