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/

Zendy Plus

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

Zendy Tools

Zendy Open

Visual retrieval system faces frequent model update and deployment. It is aheavy workload to re-extract features of the whole database every time.Featurecompatibility enables the learned new visual features to be directly comparedwith the old features stored in the database. In this way, when updating thedeployed model, we can bypass the inflexible and time-consuming featurere-extraction process. However, the old feature space that needs to becompatible is not ideal and faces the distribution discrepancy problem with thenew space caused by different supervision losses. In this work, we propose aglobal optimization Dual-Tuning method to obtain feature compatibility againstdifferent networks and losses. A feature-level prototype loss is proposed toexplicitly align two types of embedding features, by transferring globalprototype information. Furthermore, we design a component-level mutualstructural regularization to implicitly optimize the feature intrinsicstructure. Experimental results on million-scale datasets demonstrate that ourDual-Tuning is able to obtain feature compatibility without sacrificingperformance. (Our code will be avaliable athttps://github.com/yanbai1993/Dual-Tuning)

Dual-Tuning: Joint Prototype Transfer and Structure Regularization for Compatible Feature Learning