AI Paper Atlas

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

VideoDG: Generalizing Temporal Relations in Videos to Novel Domains

Zhiyu Yao, Yunbo Wang, Jianmin Wang, Philip S. Yu, Mingsheng Long

TPAMI 2021 · Label-efficient and robust learning › Domain adaptation and generalization · approach: Augmentation and mixture-of-experts generalization · 39 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 RADA adversarial augmentation to APNAPN · UCF->HMDBaccuracy (%): 54.3 -> 59.1Improved
add RADA adversarial augmentation to APNAPN · HMDB->UCFaccuracy (%): 71.4 -> 74.9Improved
apply original ADA (semantic) to NL-I3DNL-I3D · UCF->HMDBaccuracy (%): 54.0 -> 53.5Hurt
use pixel-space instead of semantic-space constraint (ADA_pixel vs RADA) (ablation)APN · UCF->HMDBaccuracy (%): RADA outperforms ADA_pixel by 2.2%Improved
add RADA instead of M-ADAAPN · UCF->HMDBaccuracy (%): 55.6 -> 59.1Improved
add random hard augmentation to APNAPN · UCF->HMDBaccuracy (%): 54.3 -> 56.1Improved
use Jigsaw puzzle instead of RADAAPN · UCF->HMDBaccuracy (%): 59.1 -> 55.2Hurt
generate adversarial data with level II and III features instead of level I only (ablation)APN + RADA · UCF->HMDBaccuracy (%): 53.6 -> 59.1Improved
add level I features to final II+III (ablation)APN + RADA · UCF->HMDBaccuracy (%): 59.1 -> 58.6Hurt
set T_max to 5 instead of 1 (ablation)RADA · UCF->HMDBaccuracy (%): 58.1 -> 59.1Improved
set T_max to 15 instead of 5 (ablation)RADA · UCF->HMDBaccuracy (%): 59.1 -> 58.1Hurt
enable robustness loss (lambda=1 vs 0) (ablation)RADA · UCF->HMDBaccuracy (%)Improved

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