Research map / Label-efficient and robust learning / Domain adaptation and generalization
MADAN: Multi-source Adversarial Domain Aggregation Network for Domain Adaptation
Sicheng Zhao, Bo Li, Pengfei Xu, Xiangyu Yue, Guiguang Ding, Kurt Keutzer
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 sub-domain aggregation discriminator (SAD) (ablation) | MADAN+ baseline · GTA+SYNTHIA to Cityscapes, FCN-VGG16 | mIoU: 30.0 -> 36.4 | Improved |
| add cross-domain cycle discriminator (CCD) (ablation) | MADAN+ baseline · GTA+SYNTHIA to Cityscapes | mIoU: 30.0 -> 33.1 | Improved |
| add dynamic semantic consistency (DSC) to SAD+CCD (ablation) | MADAN+ · GTA+SYNTHIA to Cityscapes | mIoU: 37.5 -> 40.2 | Improved |
| add feature-level alignment (FLA) (ablation) | MADAN+ · GTA+SYNTHIA to Cityscapes | mIoU: 40.2 -> 41.4 | Improved |
| add category-level alignment (CLA) and then CAG (ablation) | MADAN+ · GTA+SYNTHIA to Cityscapes | mIoU: 41.4 -> 42.8 | Improved |
| add the full set of components (ablation) | MADAN+ · GTA+SYNTHIA to BDDS | mIoU: 24.6 -> 41.6 | Improved |
| replace SC loss by dynamic SC loss (ablation) | CycleGAN · GTA to Cityscapes | mIoU: 32.7 -> 38.1 | Improved |
| replace SC loss by dynamic SC loss (ablation) | CyCADA · GTA to Cityscapes | mIoU: 38.7 -> 40.0 | Improved |
| replace SC loss by dynamic SC loss (ablation) | CycleGAN · SYNTHIA to Cityscapes | mIoU: 27.3 -> 30.5 | Improved |
| use multiple source domains with MADAN instead of best single-source DA | FCN-VGG16 · Digits-five | accuracy (%): 85.4 -> 90.9 | Improved |
| MADAN vs best baseline on Office31 | MADAN · Office31 | accuracy (%): 84.2 -> 87.2 | Improved |
| MADAN vs CyCADA on Office-Home | MADAN · Office-Home | accuracy (%): 68.3 -> 70.4 | Improved |
Later papers that compared with this method
1 later accepted paper reports MADAN as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Multi-Source Domain Adaptation With Collaborative Learning for Semantic Segmentation (CVPR 2021 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Multi-source Domain Adaptation for Semantic Segmentation (NeurIPS 2019 · similarity 0.93)
- Multi-Adversarial Domain Adaptation (AAAI 2018 · similarity 0.91)
- Adversarial Multiple Source Domain Adaptation (NeurIPS 2018 · similarity 0.89)
- MultiDIAL: Domain Alignment Layers for (Multisource) Unsupervised Domain Adaptation (TPAMI 2020 · similarity 0.89)
- Blending-Target Domain Adaptation by Adversarial Meta-Adaptation Networks (CVPR 2019 · similarity 0.89)
- Invertible Projection and Conditional Alignment for Multi-Source Blended-Target 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.
