Research map / Label-efficient and robust learning / Domain adaptation and generalization
Source-Free Domain Adaptation via Distribution Estimation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Estimate class-conditioned source distributions from target statistics calibrated by classifier anchors, then align surrogate-source and target features | SFDA-DE · Office-Home Avg. | accuracy (%): 72.8 -> 72.9 | Helped |
| Estimate class-conditioned source distributions from target statistics calibrated by classifier anchors, then align surrogate-source and target features | SFDA-DE · VisDA-2017 Avg. | accuracy (%): 85.4 -> 86.5; 1.1% | Helped |
| Calibrate the estimated source mean using anchor direction and target-feature norm rather than using the target feature mean directly (ablation) | SFDA-DE · VisDA-2017 | accuracy (%): 68.8 -> 86.5; 17.7% | Helped |
| Calibrate the estimated source mean using target-feature norm and anchor direction rather than using anchors directly (ablation) | SFDA-DE · Office-Home Avg. (three tasks) | accuracy (%): 64.0 -> 73.9 | Helped |
| Calibrate the estimated source mean using target-feature norm and anchor direction rather than using anchors directly (ablation) | SFDA-DE · VisDA-2017 | accuracy (%): 64.1 -> 86.5 | Helped |
| Re-estimate the surrogate source distributions during adaptation instead of updating them only once at the start (ablation) | SFDA-DE · VisDA-2017 | accuracy (%): 79.7 -> 86.5 | Helped |
| Use anchor-based spherical k-means pseudo-labeling rather than maximum-probability pseudo-labeling (ablation) | SFDA-DE · Office-Home Avg. (three tasks) | accuracy (%): 73.1 -> 73.9 | Helped |
| Use anchor-based spherical k-means pseudo-labeling rather than maximum-probability pseudo-labeling (ablation) | SFDA-DE · VisDA-2017 | accuracy (%): 85.3 -> 86.5 | Helped |
| Increase the covariance multiplier to gamma = 2 (ablation) | SFDA-DE · VisDA-2017 | accuracy (%): 0.2% | Helped |
| Use a larger confidence threshold to include more pseudo-labeled target data (ablation) | SFDA-DE · Office-Home Ar→Pr | accuracy | Mixed |
Closest papers by meaning
- Revisiting Source-Free Domain Adaptation: Insights into Representativeness, Generalization, and Variety (CVPR 2025 · similarity 0.92)
- Understanding and Improving Source-free Domain Adaptation from a Theoretical Perspective (CVPR 2024 · similarity 0.92)
- Dynamic Target Distribution Estimation for Source-Free Open-Set Domain Adaptation (AAAI 2025 · similarity 0.91)
- SF(DA)$^2$: Source-free Domain Adaptation Through the Lens of Data Augmentation (ICLR 2024 · similarity 0.90)
- Distinguish Then Exploit: Source-free Open Set Domain Adaptation via Weight Barcode Estimation and Sparse Label Assignment (CVPR 2025 · similarity 0.89)
- Diffusion-Driven Progressive Target Manipulation for Source-Free Domain Adaptation (NeurIPS 2025 · similarity 0.89)
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.