Research map / Label-efficient and robust learning / Domain adaptation and generalization
Deep Residual Correction Network for Partial Domain Adaptation
Shuang Li, Chi Harold Liu, Qiuxia Lin, Qi Wen, Limin Su, Gao Ming Huang, Zhengming Ding
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 |
|---|---|---|---|
| add domain-wise alignment loss L_domain (ablation) | DRCN (soft label) · Office-Home partial | accuracy (%): 59.9 -> 68.9 | Improved |
| add domain-wise alignment loss L_domain (ablation) | DRCN (soft label) · Office-31 partial | accuracy (%): 94.1 -> 94.3 | Improved |
| add weighted class-wise alignment loss L_class (ablation) | DRCN (soft label) · Office-Home partial | accuracy (%): 61.9 -> 68.9 | Improved |
| add weighted class-wise alignment loss L_class (ablation) | DRCN (soft label) · Office-31 partial | accuracy (%): 93.3 -> 94.3 | Improved |
| use learned class weights instead of all-one weights (ablation) | DRCN (soft label) · Office-Home partial | accuracy (%): 64.4 -> 68.9 | Improved |
| use learned class weights instead of all-one weights (ablation) | DRCN (soft label) · Office-31 partial | accuracy (%): 93.2 -> 94.3 | Improved |
| add residual correction block (RCB) (ablation) | DRCN (soft label) · Office-Home partial | accuracy (%): 66.3 -> 68.9 | Improved |
| add residual correction block (RCB) (ablation) | DRCN (soft label) · Office-31 partial | accuracy (%): 93.3 -> 94.3 | Improved |
| use hard labels instead of soft labels | DRCN · Office-31 partial | accuracy (%): 94.3 -> 95.9 | Improved |
| use soft labels instead of hard labels | DRCN · VisDA2017 + ImageNet-Caltech partial | accuracy (%): 70.8 -> 71.4 | Improved |
| use soft labels instead of hard labels | DRCN · Office-Home partial | accuracy (%): 68.5 -> 69.0 | Improved |
| use DRCN instead of SAN (best baseline) | DRCN · Office-Home partial | accuracy (%): 65.3 -> 69.0 | Improved |
Closest papers by meaning
- Unsupervised Domain Adaptation with Residual Transfer Networks (NeurIPS 2016 · similarity 0.90)
- Deep Reconstruction-Classification Networks for Unsupervised Domain Adaptation (ECCV 2016 · similarity 0.90)
- Residual Parameter Transfer for Deep Domain Adaptation (CVPR 2018 · similarity 0.90)
- Learning to Transfer Examples for Partial Domain Adaptation (CVPR 2019 · similarity 0.89)
- Boosting Domain Adaptation by Discovering Latent Domains (CVPR 2018 · similarity 0.88)
- Co-regularized Alignment for Unsupervised Domain Adaptation (NeurIPS 2018 · 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.
