Research map / Label-efficient and robust learning / Domain adaptation and generalization
One-Sided Unsupervised Domain Mapping
Sagie Benaim, Lior Wolf
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 pairwise distance preservation to the adversarial mapping | DiscoGAN · car2car | RMSE: 0.306 -> 0.135 | Improved |
| Add circularity to pairwise distance preservation (ablation) | Distance · car2car | RMSE: 0.135 -> 0.098 | Improved |
| Replace pairwise distances with within-image self-distances (ablation) | Distance · car2car | RMSE: 0.135 -> 0.117 | Improved |
| Add pairwise distance preservation to the adversarial mapping | DiscoGAN · car2head | RMSE: 0.137 -> 0.097 | Improved |
| Add circularity to pairwise distance preservation (ablation) | Distance · car2head | RMSE: 0.097 -> 0.273 | Hurt |
| Replace pairwise distances with within-image self-distances (ablation) | Distance · car2head | RMSE: 0.097 -> 0.197 | Hurt |
| Add pairwise distance preservation to CycleGAN | CycleGAN · SVHN to MNIST | Accuracy: 26.1% -> 26.8% | Improved |
| Add circularity to pairwise distance preservation (ablation) | Distance · SVHN to MNIST | Accuracy: 26.8% -> 18.0% | Hurt |
| Use within-image self-distances instead of pairwise distances (ablation) | Distance · SVHN to MNIST | Accuracy: 26.8% -> 25.2% | Hurt |
| Add pairwise distance preservation | DiscoGAN · CelebA attribute translations | Cosine similarity: 0.23 -> 0.32; 0.15 -> 0.24; 0.13 -> 0.42 | Improved |
| Add circularity to pairwise distance preservation (ablation) | Distance · CelebA attribute translations | Cosine similarity: 0.32 -> 0.35; 0.24 -> 0.24; 0.42 -> 0.41 | Mixed |
| Replace pairwise distances with within-image self-distances (ablation) | Distance · CelebA attribute translations | Cosine similarity: Male to Female: 0.32 -> 0.24; Blond to Black: 0.24 -> 0.24; Glasses to Without: 0.42 -> 0.34 | Mixed |
Closest papers by meaning
- Bidirectional One-Shot Unsupervised Domain Mapping (ICCV 2019 · similarity 0.87)
- Asymmetric Tri-training for Unsupervised Domain Adaptation (ICML 2017 · similarity 0.83)
- Out-Of-Domain Unlabeled Data Improves Generalization (ICLR 2024 · similarity 0.81)
- Towards Identifiable Unsupervised Domain Translation: A Diversified Distribution Matching Approach (ICLR 2024 · similarity 0.81)
- How Many Domains Suffice for Domain Generalization? A Tight Characterization via the Domain Shattering Dimension (NeurIPS 2025 · similarity 0.81)
- Active Learning of General Halfspaces: Label Queries vs Membership Queries (NeurIPS 2024 · similarity 0.81)
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.
