AI Paper Atlas

Research map / Label-efficient and robust learning / Domain adaptation and generalization

Confidence Regularized Self-Training

Yang Zou, Zhiding Yu, Xiaofeng Liu, B.V.K. Vijaya Kumar, Jinsong Wang

ICCV 2019 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Conditional distribution alignment · 885 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 MRKLD confidence regularizerCBST · VisDA17mean accuracy: 76.4 -> 77.9Improved
add LRENT label regularizerCBST · VisDA17mean accuracy: 76.4 -> 76.6Improved
add MRL2 regularizerCBST · VisDA17mean accuracy: 76.4 -> 76.2No clear effect
add MRENT regularizerCBST · VisDA17mean accuracy: 76.4 -> 76.2No clear effect
add LRENT on top of MRKLD (ablation)MRKLD · VisDA17mean accuracy: 77.9 -> 78.1Improved
combine MRKLD and LRENTCBST · Office-31mean accuracy: 85.8 -> 86.8Improved
add MRKLD regularizerCBST (DeepLabv2) · GTA5 to CityscapesmIoU: 45.9 -> 47.1Improved
add MRKLD regularizerCBST (ResNet-38) · GTA5 to CityscapesmIoU: 45.2 -> 46.8Improved
add MRKLD regularizer with spatial priorsCBST-SP (ResNet-38) · GTA5 to CityscapesmIoU: 48.1 -> 49.2Improved
add multi-scale testing (ablation)MRKLD-SP (ResNet-38) · GTA5 to CityscapesmIoU: 49.2 -> 49.8Improved
add MRKLD regularizerCBST · SYNTHIA to CityscapesmIoU: 42.6 -> 43.8Improved
vary starting portion p0 from 20 to 15 (ablation)MRL2 · Office-31 W to Aaccuracy: 72.1 -> 71.3No clear effect

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