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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Unsupervised domain adaptation (UDA) is to learn classification models thatmake predictions for unlabeled data on a target domain, given labeled data on asource domain whose distribution diverges from the target one. Mainstream UDAmethods strive to learn domain-aligned features such that classifiers trainedon the source features can be readily applied to the target ones. Althoughimpressive results have been achieved, these methods have a potential risk ofdamaging the intrinsic data structures of target discrimination, raising anissue of generalization particularly for UDA tasks in an inductive setting. Toaddress this issue, we are motivated by a UDA assumption of structuralsimilarity across domains, and propose to directly uncover the intrinsic targetdiscrimination via constrained clustering, where we constrain the clusteringsolutions using structural source regularization that hinges on the very sameassumption. Technically, we propose a hybrid model of Structurally RegularizedDeep Clustering, which integrates the regularized discriminative clustering oftarget data with a generative one, and we thus term our method as H-SRDC. Ourhybrid model is based on a deep clustering framework that minimizes theKullback-Leibler divergence between the distribution of network prediction andan auxiliary one, where we impose structural regularization by learningdomain-shared classifier and cluster centroids. By enriching the structuralsimilarity assumption, we are able to extend H-SRDC for a pixel-level UDA taskof semantic segmentation. We conduct extensive experiments on seven UDAbenchmarks of image classification and semantic segmentation. With no explicitfeature alignment, our proposed H-SRDC outperforms all the existing methodsunder both the inductive and transductive settings. We make our implementationcodes publicly available at https://github.com/huitangtang/H-SRDC.

Towards Uncovering the Intrinsic Data Structures for Unsupervised Domain Adaptation using Structurally Regularized Deep Clustering