paper-with-me

Papers

Return of Frustratingly Easy Unsupervised Video Domain Adaptation

2026-05-19 · Pengfei Wei, Yiqun Sun, Zhiqiang Xu, Yiping Ke, Lawrence B. Hsieh arxiv

Unsupervised video domain adaptation (UVDA) is a practical but under-explored problem. In this paper, we propose a frustratingly easy UVDA method, called MetaTrans. Specifically, MetaTrans adopts a concise learning objective that contains only two fundamental loss terms. Despite the simplicity of the learning objective, MetaTrans embodies an advanced UVDA idea, that is, handling the spatial and temporal divergence of cross-domain videos separately, through a subtle model architecture design. By implementing a temporal-static subtraction module, MetaTrans effectively removes spatial and temporal divergence. Extensive empirical evaluations, particularly on various cross-domain action recognition tasks, show substantial absolute adaptation performance enhancement and significantly superior relative performance gain compared with state-of-the-art UVDA baselines.

📄 PDF Abstract BibTeX arXiv:2605.19510

Code (0)

등록된 구현이 없습니다.

Tasks

Action RecognitionDomain Adaptation

Similar Papers 제목 키워드 기반

Return of Frustratingly Easy Domain Adaptation

2015-11-17 · Baochen Sun, Jiashi Feng, Kate Saenko

Unlike human learning, machine learning often fails to handle changes between training (source) and test (target) input distributions. Such domain shifts, common in practical scenarios, severely damage the performance of…

BIG-bench Machine LearningDomain AdaptationUnsupervised Domain Adaptation

Frustratingly Easy Uncertainty Estimation for Distribution Shift

2021-06-07 · Tiago Salvador, Vikram Voleti, Alexander Iannantuono, Adam Oberman

Distribution shift is an important concern in deep image classification, produced either by corruption of the source images, or a complete change, with the solution involving domain adaptation. While the primary goal is …

Domain Adaptationimage-classificationImage ClassificationUnsupervised Domain Adaptation

Decoupled Training: Return of Frustratingly Easy Multi-Domain Learning

2023-09-19 · Ximei Wang, Junwei Pan, Xingzhuo Guo, Dapeng Liu 외

Multi-domain learning (MDL) aims to train a model with minimal average risk across multiple overlapping but non-identical domains. To tackle the challenges of dataset bias and domain domination, numerous MDL approaches h…

Recommendation Systems

Deep CORAL: Correlation Alignment for Deep Domain Adaptation

2016-07-06 · Baochen Sun, Kate Saenko

Deep neural networks are able to learn powerful representations from large quantities of labeled input data, however they cannot always generalize well across changes in input distributions. Domain adaptation algorithms …

Domain AdaptationDomain GeneralizationImage ClassificationUnsupervised Domain Adaptation

Frustratingly Easy Domain Adaptation

2009-07-10 · Hal Daumé III

We describe an approach to domain adaptation that is appropriate exactly in the case when one has enough ``target'' data to do slightly better than just using only ``source'' data. Our approach is incredibly simple, easy…

Domain Adaptation