Research map / Label-efficient and robust learning / Domain adaptation and generalization
Adversarial-Learned Loss for 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 |
|---|---|---|---|
| Use ALDA's adversarially learned confusion matrix and corrected target loss | ResNet-50 domain adaptation · Office-31 | average accuracy (%): 87.8 -> 88.7 | Helped |
| Use ALDA's adversarially learned confusion matrix and corrected target loss | ResNet-50 domain adaptation · Office-Home | average accuracy (%): 65.8 -> 66.6 | Helped |
| Use ALDA | VisDA-2017 with ResNet-101 · VisDA-2017 validation | average accuracy (%): 76.4 -> 77.8 | Helped |
| Use ALDA | VisDA-2017 with ResNet-50 · VisDA-2017 validation | average accuracy (%): 70.0 -> 76.5 | Helped |
| Use ALDA with pseudo-label threshold δ=0.6 | Digit domain adaptation · Digits | average accuracy (%): 95.9 -> 97.6 | Helped |
| Add discriminator regularization L_Reg (ablation) | ALDA · Office-31 | average accuracy (%): 87.0 -> 88.7 | Helped |
| Add corrected target loss L_T (ablation) | ALDA · Office-31 | average accuracy (%): 87.7 -> 88.7 | Helped |
| Replace self-training pseudo-label loss with corrected target loss L_T (ablation) | ALDA+ST · Office-31 | average accuracy (%): 87.8 -> 88.7 | Helped |
| Use unhinged loss rather than cross-entropy in corrected target loss (ablation) | ALDA · Office-31 | average accuracy (%): 87.6 -> 88.7 | Helped |
| Lower the pseudo-label threshold from δ=0.9 to δ=0.6 (ablation) | ALDA · Digits | average accuracy (%): 96.2 -> 97.6 | Helped |
| Lower the pseudo-label threshold from δ=0.8 to δ=0.6 (ablation) | ALDA · Digits | average accuracy (%): 97.0 -> 97.6 | Helped |
| Lower the pseudo-label threshold from δ=0.0 to δ=0.6 (ablation) | ALDA · Digits | average accuracy (%): 96.8 -> 97.6 | Helped |
Closest papers by meaning
- Adversarial Discriminative Domain Adaptation (CVPR 2017 · similarity 0.90)
- Blending-Target Domain Adaptation by Adversarial Meta-Adaptation Networks (CVPR 2019 · similarity 0.90)
- Transferable Adversarial Training: A General Approach to Adapting Deep Classifiers (ICML 2019 · similarity 0.90)
- Bridging Theory and Algorithm for Domain Adaptation (ICML 2019 · similarity 0.90)
- Domain Adaptation with Conditional Distribution Matching and Generalized Label Shift (NeurIPS 2020 · similarity 0.89)
- Toward Improving Robustness and Accuracy in Unsupervised 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.