English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Multimedia applications are often associated with cross-domain knowledgetransfer, where Unsupervised Domain Adaptation (UDA) can be used to reduce thedomain shifts. Open Set Domain Adaptation (OSDA) aims to transfer knowledgefrom a well-labeled source domain to an unlabeled target domain under theassumption that the target domain contains unknown classes. Existing OSDAmethods consistently lay stress on the covariate shift, ignoring the potentiallabel shift problem. The performance of OSDA methods degrades drastically underintra-domain class imbalance and inter-domain label shift. However, littleattention has been paid to this issue in the community. In this paper, theImbalanced Open Set Domain Adaptation (IOSDA) is explored where the covariateshift, label shift and category mismatch exist simultaneously. To alleviate thenegative effects raised by label shift in OSDA, we propose Open-setMoving-threshold Estimation and Gradual Alignment (OMEGA) - a novelarchitecture that improves existing OSDA methods on class-imbalanced data.Specifically, a novel unknown-aware target clustering scheme is proposed toform tight clusters in the target domain to reduce the negative effects oflabel shift and intra-domain class imbalance. Furthermore, moving-thresholdestimation is designed to generate specific thresholds for each target samplerather than using one for all. Extensive experiments on IOSDA, OSDA and OPDAbenchmarks demonstrate that our method could significantly outperform existingstate-of-the-arts. Code and data are available athttps://github.com/mendicant04/OMEGA.

Imbalanced Open Set Domain Adaptation via Moving-threshold Estimation and Gradual Alignment