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

Deep supervised hashing has emerged as an influential solution to large-scalesemantic image retrieval problems in computer vision. In the light of recentprogress, convolutional neural network based hashing methods typically seekpair-wise or triplet labels to conduct the similarity preserving learning.However, complex semantic concepts of visual contents are hard to capture bysimilar/dissimilar labels, which limits the retrieval performance. Generally,pair-wise or triplet losses not only suffer from expensive training costs butalso lack in extracting sufficient semantic information. In this regard, wepropose a novel deep supervised hashing model to learn more compact class-levelsimilarity preserving binary codes. Our deep learning based model is motivatedby deep metric learning that directly takes semantic labels as supervisedinformation in training and generates corresponding discriminant hashing code.Specifically, a novel cubic constraint loss function based on Gaussiandistribution is proposed, which preserves semantic variations while penalizesthe overlap part of different classes in the embedding space. To address thediscrete optimization problem introduced by binary codes, a two-stepoptimization strategy is proposed to provide efficient training and avoid theproblem of gradient vanishing. Extensive experiments on four large-scalebenchmark databases show that our model can achieve the state-of-the-artretrieval performance. Moreover, when training samples are limited, our methodsurpasses other supervised deep hashing methods with non-negligible margins.

Deep Class-Wise Hashing: Semantics-Preserving Hashing via Class-wise Loss