Research map / Label-efficient and robust learning / Domain adaptation and generalization
Feature-Level Domain Adaptation
Wouter M. Kouw, Laurens J.P. van der Maaten, Jesse H. Krijthe, Marco Loog
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 |
|---|---|---|---|
| perturb estimated transfer parameter theta_1 by +0.1, +0.2, +0.3 (ablation) | FLDA-Q · Synthetic Poisson | classification error: .138 -> .145 -> .149 -> .150 | Hurt |
| perturb estimated transfer parameter theta_1 by +0.1, +0.2, +0.3 (ablation) | FLDA-L · Synthetic Poisson | classification error: .139 -> .140 -> .142 -> .146 | Hurt |
| adapt with dropout transfer model (FLDA-Q) instead of naive source least squares | S-LS · Synthetic Bernoulli | classification error: 0.400 -> 0.300 | Improved |
| adapt with dropout transfer model (FLDA-Q) instead of naive source least squares | S-LS · Synthetic Poisson | classification error: 0.181 -> 0.099 | Improved |
| adapt with FLDA-L instead of naive source logistic regression | S-LR · Synthetic Poisson | classification error: 0.170 -> 0.084 | Improved |
| adapt with FLDA-L instead of naive source logistic regression | S-LR · Synthetic Bernoulli | classification error: 0.248 -> 0.301 | Hurt |
| use FLDA-Q with few training samples versus naive classifiers | S-LS and T-LS · Synthetic Poisson learning curves | classification error | Improved |
| apply FLDA (L and Q) to missing-data-at-test-time | S-LR · UCI Heart Disease | classification error: .338 -> .203 | Improved |
| apply FLDA-L to missing-data-at-test-time | S-LR · UCI Hepatitis | classification error: .493 -> .200 | Improved |
| apply FLDA-L | S-LR · MNIST/USPS digits M->U | classification error: .747 -> .678 | No clear effect |
| apply FLDA-Q | S-LR · MNIST/USPS digits U->M | classification error: .770 -> .640 | Improved |
| use FLDA-L instead of FLDA-Q | FLDA · Amazon sentiment (B->D) | classification error: .303 -> .166 | Improved |
Closest papers by meaning
- A Review of Domain Adaptation without Target Labels (TPAMI 2019 · similarity 0.90)
- Domain Adaptation with Conditional Transferable Components (ICML 2016 · similarity 0.88)
- Unsupervised Domain Adaptation With Distribution Matching Machines (AAAI 2018 · similarity 0.88)
- Open-Set Heterogeneous Domain Adaptation: Theoretical Analysis and Algorithm (AAAI 2025 · similarity 0.86)
- Understanding and Improving Source-free Domain Adaptation from a Theoretical Perspective (CVPR 2024 · similarity 0.86)
- A Separation and Alignment Framework for Black-Box Domain Adaptation (AAAI 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.
