Research map / Label-efficient and robust learning / Domain adaptation and generalization
Cooperative and Adversarial Learning: Co-enhancing Discriminability and Transferability in 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 CALE instead of CST | Office-Home with ResNet-50 · Office-Home | accuracy (Avg): 73.0 -> 74.8 | Helped |
| Use CALE instead of CDTrans | Office-Home with DeiT-Base · Office-Home | accuracy (Avg): 80.5 -> 81.5 | Helped |
| Use CALE instead of CST | VisDA-2017 with ResNet-101 · VisDA-2017 | accuracy (Avg): 84.9 -> 87.9 | Helped |
| Use CALE instead of CDTrans | VisDA-2017 with DeiT-Base · VisDA-2017 | accuracy (Avg): 88.4 -> 90.8 | Helped |
| Add transferability regularization (ablation) | Office-Home ResNet-50 UDA model with transferability loss · Office-Home | accuracy (Avg): 63.8 -> 66.3 | Helped |
| Add discriminability regularization to transferability regularization (ablation) | Office-Home ResNet-50 UDA model · Office-Home | accuracy (Avg): 66.3 -> 71.4 | Helped |
| Use cooperative learning rather than the corresponding regularization variant without it (ablation) | Office-Home ResNet-50 UDA model · Office-Home | accuracy (Avg): 71.4 -> 72.3 | Helped |
| Add adversarial learning to cooperative learning (ablation) | Office-Home ResNet-50 UDA model · Office-Home | accuracy (Avg): 72.3 -> 73.9 | Helped |
| Add self-training to CALE (ablation) | Office-Home ResNet-50 UDA model · Office-Home | accuracy (Avg): 73.9 -> 74.8 | Helped |
| Remove self-training (ablation) | CALE with ResNet-50 · Office-Home | accuracy (Avg): 74.8 -> 73.9 | Hurt |
Closest papers by meaning
- Transferable Adversarial Training: A General Approach to Adapting Deep Classifiers (ICML 2019 · similarity 0.90)
- Transferability vs. Discriminability: Batch Spectral Penalization for Adversarial Domain Adaptation (ICML 2019 · similarity 0.90)
- A DIRT-T Approach to Unsupervised Domain Adaptation (ICLR 2018 · similarity 0.90)
- Blending-Target Domain Adaptation by Adversarial Meta-Adaptation Networks (CVPR 2019 · similarity 0.89)
- Bridging Theory and Algorithm for Domain Adaptation (ICML 2019 · similarity 0.89)
- Reusing the Task-Specific Classifier as a Discriminator: Discriminator-Free Adversarial Domain Adaptation (CVPR 2022 · similarity 0.89)
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.