AI Paper Atlas

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

Partial Video Domain Adaptation With Partial Adversarial Temporal Attentive Network

Yuecong Xu, Jianfei Yang, Haozhi Cao, Zhenghua Chen, Qi Li, Kezhi Mao

ICCV 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Adversarial domain adaptation · 29 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 label attention weights (ablation)PATAN · U-14->H-7top-1 accuracy: 71.43 -> 73.81Improved
add label attention weights (ablation)PATAN · H-14->U-7top-1 accuracy: 85.34 -> 89.85Improved
add local temporal weights to class weights (ablation)PATAN · U-14->H-7top-1 accuracy: 70.47 -> 73.81Improved
add local temporal weights to class weights (ablation)PATAN · H-14->U-7top-1 accuracy: 84.21 -> 89.85Improved
apply class weights to source classifiers (ablation)PATAN · U-14->H-7top-1 accuracy: 69.52 -> 73.81Improved
apply class weights to source classifiers (ablation)PATAN · H-14->U-7top-1 accuracy: 82.71 -> 89.85Improved
apply class weights to domain discriminators (ablation)PATAN · U-14->H-7top-1 accuracy: 67.14 -> 73.81Improved
apply class weights to domain discriminators (ablation)PATAN · H-14->U-7top-1 accuracy: 81.58 -> 89.85Improved
use UDA/VUDA methods (DANN, TA3N, MK-MMD, MCD, MDD) instead of source-only TRNTRN · six PVDA settingstop-1 accuracyHurt
use PATAN instead of PADAPADA · six PVDA settingstop-1 accuracy: 11.27% average relative improvementImproved
reduce number of target classesDANN, PADA, ETN · UCF-HMDB partialtarget accuracyHurt
reduce number of target classesPATAN · UCF-HMDB partialtarget accuracyNo 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