AI Paper Atlas

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

IJCV 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Adaptation for semantic segmentation · 82 citations (OpenAlex)

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 changedWhereMeasuredEffect
add sub-domain aggregation discriminator (SAD) (ablation)MADAN+ baseline · GTA+SYNTHIA to Cityscapes, FCN-VGG16mIoU: 30.0 -> 36.4Improved
add cross-domain cycle discriminator (CCD) (ablation)MADAN+ baseline · GTA+SYNTHIA to CityscapesmIoU: 30.0 -> 33.1Improved
add dynamic semantic consistency (DSC) to SAD+CCD (ablation)MADAN+ · GTA+SYNTHIA to CityscapesmIoU: 37.5 -> 40.2Improved
add feature-level alignment (FLA) (ablation)MADAN+ · GTA+SYNTHIA to CityscapesmIoU: 40.2 -> 41.4Improved
add category-level alignment (CLA) and then CAG (ablation)MADAN+ · GTA+SYNTHIA to CityscapesmIoU: 41.4 -> 42.8Improved
add the full set of components (ablation)MADAN+ · GTA+SYNTHIA to BDDSmIoU: 24.6 -> 41.6Improved
replace SC loss by dynamic SC loss (ablation)CycleGAN · GTA to CityscapesmIoU: 32.7 -> 38.1Improved
replace SC loss by dynamic SC loss (ablation)CyCADA · GTA to CityscapesmIoU: 38.7 -> 40.0Improved
replace SC loss by dynamic SC loss (ablation)CycleGAN · SYNTHIA to CityscapesmIoU: 27.3 -> 30.5Improved
use multiple source domains with MADAN instead of best single-source DAFCN-VGG16 · Digits-fiveaccuracy (%): 85.4 -> 90.9Improved
MADAN vs best baseline on Office31MADAN · Office31accuracy (%): 84.2 -> 87.2Improved
MADAN vs CyCADA on Office-HomeMADAN · Office-Homeaccuracy (%): 68.3 -> 70.4Improved

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy