Research map / Label-efficient and robust learning / Domain adaptation and generalization
Addressing Parameter Choice Issues in Unsupervised Domain Adaptation by Aggregation
Marius-Constantin Dinu, Markus Holzleitner, Maximilian Beck, Hoan Duc Nguyen, Andrea Huber, Hamid Eghbal-zadeh, Bernhard A. Moser, Sergei Pereverzyev, Sepp Hochreiter, Werner Zellinger
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 |
|---|---|---|---|
| Aggregate candidate models using importance-weighted least squares | Unsupervised domain adaptation parameter choice · Amazon Reviews, Sleep-EDF, UCI-HAR, HHAR, WISDM, and two other datasets | target accuracy | Improved |
| Use importance-weighted least squares aggregation instead of IWV or DEV model selection | Unsupervised domain adaptation parameter choice · All seven datasets | target accuracy | Improved |
| Aggregate candidate models with IWA instead of heuristic aggregation | Unsupervised domain adaptation parameter choice · Seven datasets | target accuracy | Mixed |
| Use target majority voting, target majority voting regression, target confidence average regression, or source-only regression instead of IWV/DEV | Unsupervised domain adaptation parameter choice · Seven datasets | target accuracy | Improved |
| Use linear aggregation rather than selecting a single candidate model | Candidate models for unsupervised domain adaptation · UCI-HAR | target accuracy: 0.850 -> 0.835 | Hurt |
| Use linear aggregation rather than selecting a single candidate model | Candidate models for unsupervised domain adaptation · HHAR | target accuracy: 0.784 -> 0.787 | Improved |
| Use linear aggregation rather than selecting a single candidate model | Candidate models for unsupervised domain adaptation · Amazon Reviews | target accuracy: 0.781 -> 0.788 | No clear effect |
| Use linear aggregation rather than selecting a single candidate model | Candidate models for unsupervised domain adaptation · Sleep-EDF | target accuracy: 0.712 -> 0.737 | Improved |
| Use linear aggregation rather than selecting a single candidate model | Candidate models for unsupervised domain adaptation · WISDM | target accuracy: 0.771 -> 0.764 | Hurt |
Closest papers by meaning
- Towards Reliable Model Selection for Unsupervised Domain Adaptation: An Empirical Study and A Certified Baseline (NeurIPS 2024 · similarity 0.89)
- On Balancing Bias and Variance in Unsupervised Multi-Source-Free Domain Adaptation (ICML 2023 · similarity 0.88)
- Prior Knowledge Guided Unsupervised Domain Adaptation (ECCV 2022 · similarity 0.88)
- UDA-Bench: Revisiting Common Assumptions in Unsupervised Domain Adaptation Using a Standardized Framework (ECCV 2024 · similarity 0.88)
- Tune It the Right Way: Unsupervised Validation of Domain Adaptation via Soft Neighborhood Density (ICCV 2021 · similarity 0.88)
- Can We Evaluate Domain Adaptation Models Without Target-Domain Labels? (ICLR 2024 · similarity 0.87)
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.
