Research map / Label-efficient and robust learning / Domain adaptation and generalization
Information-Theoretic Analysis of Unsupervised 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 |
|---|---|---|---|
| Add gradient penalty to KL (ablation) | KL · RotatedMNIST average | accuracy: 86.4 -> 91.2 | Helped |
| Add gradient penalty to KL (ablation) | KL · Digits average | accuracy: 96.0 -> 96.8 | Helped |
| Add gradient penalty to KL (ablation) | KL · RotatedMNIST 30° | accuracy: 97.1±0.2 -> 96.9±0.1 | Hurt |
| Add gradient penalty to KL (ablation) | KL · RotatedMNIST 45°, 60°, and 75° | accuracy: 93.4±0.8 -> 95.0±0.6; 75.5±2.4 -> 88.0±8.1; 68.1±1.8 -> 78.1±2.5 | Helped |
| Add label-information control to KL (ablation) | KL · RotatedMNIST average | accuracy: 86.4 -> 89.6 | Helped |
| Add label-information control to KL (ablation) | KL · Digits average | accuracy: 96.0 -> 96.5 | Helped |
| Add label-information control to KL (ablation) | KL · RotatedMNIST 75° | accuracy: 68.1±1.8 -> 73.6±4.0 | Helped |
| Add label-information control to KL (ablation) | KL · Digits M→U | accuracy: 98.2±0.2 -> 98.9±0.1 | Helped |
| Add label-information control to KL (ablation) | KL · Digits U→M and S→M | accuracy: 97.3±0.5 -> 97.7±0.1; 92.5±0.9 -> 93.0±0.3 | Mixed |
| Add gradient penalty to ERM (ablation) | ERM · RotatedMNIST average | accuracy: 58.4 -> 61.2 | Helped |
| Add gradient penalty to ERM (ablation) | ERM · Digits average | accuracy: 64.6 -> 77.5 | Helped |
Closest papers by meaning
- Domain Adaptation with Adaptive $f$-Divergence: Tighter Variational Representation and Generalization Bounds (ICML 2026 · similarity 0.90)
- On $f$-Divergence Principled Domain Adaptation: An Improved Framework (NeurIPS 2024 · similarity 0.87)
- Unknown Domain Inconsistency Minimization for Domain Generalization (ICLR 2024 · similarity 0.87)
- Understanding and Improving Source-free Domain Adaptation from a Theoretical Perspective (CVPR 2024 · similarity 0.87)
- Can We Evaluate Domain Adaptation Models Without Target-Domain Labels? (ICLR 2024 · similarity 0.86)
- Towards Generalization beyond Pointwise Learning: A Unified Information-theoretic Perspective (ICML 2024 · 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.