Research map / Label-efficient and robust learning / Domain adaptation and generalization
Forget More to Learn More: Domain-specific Feature Unlearning for Semi-supervised and Unsupervised Domain Adaptation
Hritam Basak, Zhaozheng Yin
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 Gaussian-guided latent alignment (KL loss) (ablation) | SSDA framework · Office-Home and DomainNet SSDA | accuracy (%): 70.4 -> 74.8 (Office-Home 1-shot) | Improved |
| add unlearning loss that maximizes uncertainty on reconstructed images (ablation) | SSDA framework · Office-Home and DomainNet SSDA | accuracy (%): 68.5 -> 74.8 (Office-Home 1-shot) | Improved |
| add reconstruction loss to keep structural similarity (ablation) | SSDA framework · Office-Home and DomainNet SSDA | accuracy (%): 71.1 -> 74.8 (Office-Home 1-shot) | Improved |
| add Gaussian-guided latent alignment for UDA (ablation) | UDA framework (ViT) · Office-Home and DomainNet UDA | accuracy (%): 82.2 -> 88.8 (Office-Home) | Improved |
| add unlearning loss for UDA (ablation) | UDA framework (ViT) · Office-Home and DomainNet UDA | accuracy (%): 80.8 -> 88.8 (Office-Home) | Improved |
| add reconstruction loss for UDA (ablation) | UDA framework (ViT) · Office-Home and DomainNet UDA | accuracy (%): 84.1 -> 88.8 (Office-Home) | Improved |
| use transformer backbone instead of ResNet | UDA framework · Office-Home and DomainNet UDA | accuracy (%): 74.1 -> 88.8 (ResNet vs ViT, Office-Home) | Improved |
| increase unlearning weight gamma relative to reconstruction weight beta (ablation) | reconstruction network · visual reconstruction grid | retained structure vs domain-specific features | Mixed |
| apply method with ViT on Office-Home UDA compared to PMTrans-ViT | UDA framework · Office-Home UDA (ViT) | avg accuracy (%): 88.8 -> 88.8 | No clear effect |
Closest papers by meaning
- Attract, Perturb, and Explore: Learning a Feature Alignment Network for Semi-supervised Domain Adaptation (ECCV 2020 · similarity 0.91)
- Semi-Supervised Domain Adaptation With Source Label Adaptation (CVPR 2023 · similarity 0.91)
- Universal Semi-Supervised Domain Adaptation by Mitigating Common-Class Bias (CVPR 2024 · similarity 0.90)
- Open-Set Heterogeneous Domain Adaptation: Theoretical Analysis and Algorithm (AAAI 2025 · similarity 0.89)
- SemiDAViL: Semi-supervised Domain Adaptation with Vision-Language Guidance for Semantic Segmentation (CVPR 2025 · similarity 0.88)
- Understanding and Improving Source-free Domain Adaptation from a Theoretical Perspective (CVPR 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.
