Research map / Label-efficient and robust learning / Domain adaptation and generalization
On Balancing Bias and Variance in Unsupervised Multi-Source-Free Domain Adaptation
Read the paperCheck 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 |
|---|---|---|---|
| uniform source weights replaced by learned domain aggregation weights (add learned weights) (ablation) | proposed MSFDA · Digits-Five | Avg accuracy: +1.2 (96.8 vs 95.6) | Helped |
| uniform source weights replaced by learned domain aggregation weights (add learned weights) (ablation) | proposed MSFDA · Office-31 | Avg accuracy: +0.6 (92.0 vs 91.4) | Helped |
| add pseudo-label denoising (prototype-based) (ablation) | proposed MSFDA · Digits-Five | Avg accuracy: +0.2 (96.8 vs 96.6) | Helped |
| add pseudo-label denoising (prototype-based) (ablation) | proposed MSFDA · Office-31 | Avg accuracy: +0.5 (92.0 vs 91.5) | Helped |
| apply information-maximization loss only on labeled subset instead of all target data (ablation) | proposed MSFDA · Digits-Five | Avg accuracy: +1.0 (96.8 vs 95.8) | Helped |
| apply information-maximization loss only on labeled subset instead of all target data (ablation) | proposed MSFDA · Office-31 | Avg accuracy: +1.4 (92.0 vs 90.6) | Helped |
| add joint adversarial feature alignment (ablation) | proposed MSFDA · Digits-Five | Avg accuracy: +0.1 (96.8 vs 96.7) | Helped |
| add joint adversarial feature alignment (ablation) | proposed MSFDA · Office-31 | Avg accuracy: +0.2 (92.0 vs 91.8) | Helped |
| add mixup augmentation to enlarge labeled subset (ablation) | proposed MSFDA · Digits-Five | Avg accuracy: +0.5 (96.8 vs 96.3) | Helped |
| add mixup augmentation to enlarge labeled subset (ablation) | proposed MSFDA · Office-31 | Avg accuracy: +0.2 (92.0 vs 91.8) | Helped |
| surrogate selection (confidence query) replaced by oracle selection | proposed MSFDA · Office-Home | Avg accuracy: +12.9 (90.0 vs 77.1) | Helped |
| proposed surrogate method vs prior SOTA on Office-Home | proposed MSFDA · Office-Home | Avg accuracy: +0.9 over CAiDA (77.1 vs 76.2) | Helped |
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 0.91)
- Unsupervised Multi-Source Domain Adaptation Without Access to Source Data (CVPR 2021 · similarity 0.91)
- Universal Semi-Supervised Domain Adaptation by Mitigating Common-Class Bias (CVPR 2024 · similarity 0.89)
- Pseudo-Labeling for Unsupervised Domain Adaptation with Kernel GLMs (NeurIPS 2026 · similarity 0.89)
- Invertible Projection and Conditional Alignment for Multi-Source Blended-Target Domain Adaptation (AAAI 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.