Research map / Label-efficient and robust learning / Domain adaptation and generalization
Divide to Adapt: Mitigating Confirmation Bias for Domain Adaptation of Black-Box Predictors
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 |
|---|---|---|---|
| Use BETA instead of the strongest reported DABP baseline | Black-box domain adaptation · VisDA-17 | Per-class accuracy: +7.0% | Helped |
| Use BETA instead of the strongest reported DABP baseline | Black-box domain adaptation · Office-31 | average accuracy: +1.8% | Helped |
| Use BETA instead of the strongest reported DABP baseline | Black-box domain adaptation · Office-Home | average accuracy: +2.4% | Helped |
| Add domain-division semi-supervised learning (ablation) | Source-only model on Office-Home · Office-Home hard tasks | H. Avg. accuracy: +11.4% | Helped |
| Add mutually-teaching twin networks (ablation) | Domain-division semi-supervised learning · Office-Home hard tasks | H. Avg. accuracy: +0.6% | Helped |
| Add mutual information maximization (ablation) | Domain-division semi-supervised learning with MTN · Office-Home hard tasks | H. Avg. accuracy: +0.9% | Helped |
| Add knowledge distillation (ablation) | Domain-division semi-supervised learning with MTN and mutual information · Office-Home hard tasks | H. Avg. accuracy: +1.1% | Helped |
| Add adversarial alignment between the easy and hard subdomains (ablation) | BETA without the adversarial regularizer · Office-Home hard tasks | H. Avg. accuracy | Helped |
| Increase the domain-division threshold τ from 0.3 to 0.9 (ablation) | BETA · Office-Home (Cl→Pr) | accuracy: 78.2% to 78.8% | No clear effect |
| Use a larger domain-division threshold τ (ablation) | BETA with mutually-teaching twin networks · Office-Home (Cl→Pr) | accuracy: 78.2% to 78.8% | Mixed |
| Use BETA instead of the second-best method on hard transfer tasks | Black-box domain adaptation · VisDA-17 hard tasks | H. Avg. accuracy: +6.7% | Helped |
Closest papers by meaning
- Reviewing the Forgotten Classes for Domain Adaptation of Black-Box Predictors (AAAI 2024 · similarity 0.93)
- Revisiting the Domain Shift and Sample Uncertainty in Multi-source Active Domain Transfer (CVPR 2024 · similarity 0.90)
- A Separation and Alignment Framework for Black-Box Domain Adaptation (AAAI 2024 · similarity 0.89)
- Revisiting Source-Free Domain Adaptation: Insights into Representativeness, Generalization, and Variety (CVPR 2025 · similarity 0.89)
- ADU: Adaptive Detection of Unknown Categories in Black-Box Domain Adaptation (CVPR 2025 · similarity 0.89)
- Gradient-Based Sample Selection for Black-Box Universal Domain Adaptation (AAAI 2025 · similarity 0.88)
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.