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

Unsupervised domain adaptation has been widely adopted to generalize modelsfor unlabeled data in a target domain, given labeled data in a source domain,whose data distributions differ from the target domain. However, existing worksare inapplicable to face recognition under privacy constraints because theyrequire sharing sensitive face images between two domains. To address thisproblem, we propose a novel unsupervised federated face recognition approach(FedFR). FedFR improves the performance in the target domain by iterativelyaggregating knowledge from the source domain through federated learning. Itprotects data privacy by transferring models instead of raw data betweendomains. Besides, we propose a new domain constraint loss (DCL) to regularizesource domain training. DCL suppresses the data volume dominance of the sourcedomain. We also enhance a hierarchical clustering algorithm to predict pseudolabels for the unlabeled target domain accurately. To this end, FedFR forms anend-to-end training pipeline: (1) pre-train in the source domain; (2) predictpseudo labels by clustering in the target domain; (3) conductdomain-constrained federated learning across two domains. Extensive experimentsand analysis on two newly constructed benchmarks demonstrate the effectivenessof FedFR. It outperforms the baseline and classic methods in the target domainby over 4% on the more realistic benchmark. We believe that FedFR will shedlight on applying federated learning to more computer vision tasks underprivacy constraints.

Towards Unsupervised Domain Adaptation for Deep Face Recognition under Privacy Constraints via Federated Learning