AI Paper Atlas

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

Geometric Understanding of Discriminability and Transferability for Visual Domain Adaptation

You-Wei Luo, Chuan-Xian Ren, Xiaolin Xu, Qingshan Liu

TPAMI 2024 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Geometric alignment and synthetic-to-real adaptation · 13 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 geometry-oriented loss L_GO (ablation)GOAL · Image-CLEF target representationLDA discriminant value: 0.095 -> 1.194Improved
add geometry-oriented loss L_GO (ablation)GOAL · Image-CLEF target representationintra-class scatter: 0.912 -> 0.450Improved
add geometry-oriented loss L_GO (ablation)GOAL · Image-CLEF target representationprincipal angle cosine: 0.670 -> 0.994Improved
add geometry-oriented loss L_GO (ablation)GOAL · Image-CLEF target representationclass-wise principal angle cosine: 0.707 -> 0.909Improved
add CDAN module to GOALGOAL · Office-31accuracy: 93.8 -> 94.1Improved
add CDAN module to GOALGOAL · Image-CLEFaccuracy: 91.3 -> 91.5Improved
add CDAN module to GOALGOAL · VisDA-2017accuracy: 83.7 -> 85.0Improved
add CDAN module to GOALGOAL · DomainNetaccuracy: 33.3 -> 34.9Improved
add CDAN module to GOALGOAL · Office-Homeaccuracy: 68.6 -> 69.7Improved
set lambda_TB larger than lambda_DB (ablation)GOAL · Office-Home, Office-31, Image-CLEF, VisDA-2017accuracyHurt
keep lambda within (0, 1+sqrt(2)] with balanced weights (ablation)GOAL · Office-Home, Office-31, Image-CLEF, VisDA-2017accuracyImproved
add warm-up training before geometric learning (ablation)GOAL · Office-Home pseudo-labelspseudo-label accuracy: about 10% ~ 14% gainImproved

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