Research map / Label-efficient and robust learning / Domain adaptation and generalization
Conditional Adversarial 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 entropy conditioning (ablation) | CDAN · Office-31 (AlexNet), average across transfer tasks | accuracy: 77.0 -> 77.7 | Helped |
| Add entropy conditioning (ablation) | CDAN · Office-31 (ResNet-50), average across transfer tasks | accuracy: 86.6 -> 87.7 | Helped |
| Add entropy conditioning (ablation) | CDAN · ImageCLEF-DA (AlexNet), average across transfer tasks | accuracy: 79.8 -> 80.3 | Helped |
| Add entropy conditioning (ablation) | CDAN · ImageCLEF-DA (ResNet-50), average across transfer tasks | accuracy: 87.1 -> 87.7 | Helped |
| Add entropy conditioning (ablation) | CDAN · Office-Home (AlexNet), average across transfer tasks | accuracy: 49.9 -> 51.0 | Helped |
| Add entropy conditioning (ablation) | CDAN · Office-Home (ResNet-50), average across transfer tasks | accuracy: 63.8 -> 65.8 | Helped |
| Add entropy conditioning (ablation) | CDAN · Digits, average across transfer tasks | accuracy: 93.1 -> 94.3 | Helped |
| Add entropy conditioning (ablation) | CDAN · VisDA-2017 Synthetic → Real | accuracy: 66.8 -> 70.0 | Helped |
| Use Gaussian random sampling for the multilinear map (ablation) | CDAN+E · Office-31 (ResNet-50), average across transfer tasks | accuracy: 87.7 -> 86.4 | Hurt |
| Use Uniform random sampling for the multilinear map (ablation) | CDAN+E · Office-31 (ResNet-50), average across transfer tasks | accuracy: 87.7 -> 87.0 | Hurt |
| Replace multilinear conditioning with concatenation of features and classifier predictions (ablation) | CDAN · Office-31 A → W and A → D (ResNet-50) | accuracy | Hurt |
| Use separate domain discriminators on feature and classifier layers instead of joint conditioning (ablation) | CDAN · Office-31 A → W (ResNet-50) | accuracy | Hurt |
Closest papers by meaning
- Domain Conditioned Adaptation Network (AAAI 2020 · similarity 0.90)
- Adversarial Discriminative Domain Adaptation (CVPR 2017 · similarity 0.90)
- Multi-Adversarial Domain Adaptation (AAAI 2018 · similarity 0.90)
- Transferable Adversarial Training: A General Approach to Adapting Deep Classifiers (ICML 2019 · similarity 0.90)
- Reusing the Task-Specific Classifier as a Discriminator: Discriminator-Free Adversarial Domain Adaptation (CVPR 2022 · similarity 0.90)
- CASUAL: Conditional Support Alignment for Domain Adaptation with Label Shift (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.