Deep learning to ternary hash codes by continuation

Abstract Recently, ithas been observed that {0,±1}‐ternary codes, which are simply generated from deep features by hard thresholding, tend to outperform {−1,1}‐binary codes in image retrieval. To obtain better ternary codes, the authors for the first time propose to jointly learn the features with t...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Mingrui Chen, Weiyu Li, Weizhi Lu
Formato: article
Lenguaje:EN
Publicado: Wiley 2021
Materias:
Acceso en línea:https://doaj.org/article/a5c841e2503147e9bb25a3892e2fa9f6
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:a5c841e2503147e9bb25a3892e2fa9f6
record_format dspace
spelling oai:doaj.org-article:a5c841e2503147e9bb25a3892e2fa9f62021-11-19T05:42:54ZDeep learning to ternary hash codes by continuation1350-911X0013-519410.1049/ell2.12317https://doaj.org/article/a5c841e2503147e9bb25a3892e2fa9f62021-11-01T00:00:00Zhttps://doi.org/10.1049/ell2.12317https://doaj.org/toc/0013-5194https://doaj.org/toc/1350-911XAbstract Recently, ithas been observed that {0,±1}‐ternary codes, which are simply generated from deep features by hard thresholding, tend to outperform {−1,1}‐binary codes in image retrieval. To obtain better ternary codes, the authors for the first time propose to jointly learn the features with the codes by appending a smoothed function to the networks. During training, the function could evolve into a non‐smoothed ternary function by a continuation method, and then generate ternary codes. The method circumvents the difficulty of directly training discrete functions and reduces the quantization errors of ternary codes. Experiments show that the proposed joint learning indeed could produce better ternary codes. For the first time, the authors propose to generate ternary hash codes by jointly learning the codes with deep features via a continuation method. Experiments show that the proposed method outperforms existing methods.Mingrui ChenWeiyu LiWeizhi LuWileyarticleElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENElectronics Letters, Vol 57, Iss 24, Pp 925-926 (2021)
institution DOAJ
collection DOAJ
language EN
topic Electrical engineering. Electronics. Nuclear engineering
TK1-9971
spellingShingle Electrical engineering. Electronics. Nuclear engineering
TK1-9971
Mingrui Chen
Weiyu Li
Weizhi Lu
Deep learning to ternary hash codes by continuation
description Abstract Recently, ithas been observed that {0,±1}‐ternary codes, which are simply generated from deep features by hard thresholding, tend to outperform {−1,1}‐binary codes in image retrieval. To obtain better ternary codes, the authors for the first time propose to jointly learn the features with the codes by appending a smoothed function to the networks. During training, the function could evolve into a non‐smoothed ternary function by a continuation method, and then generate ternary codes. The method circumvents the difficulty of directly training discrete functions and reduces the quantization errors of ternary codes. Experiments show that the proposed joint learning indeed could produce better ternary codes. For the first time, the authors propose to generate ternary hash codes by jointly learning the codes with deep features via a continuation method. Experiments show that the proposed method outperforms existing methods.
format article
author Mingrui Chen
Weiyu Li
Weizhi Lu
author_facet Mingrui Chen
Weiyu Li
Weizhi Lu
author_sort Mingrui Chen
title Deep learning to ternary hash codes by continuation
title_short Deep learning to ternary hash codes by continuation
title_full Deep learning to ternary hash codes by continuation
title_fullStr Deep learning to ternary hash codes by continuation
title_full_unstemmed Deep learning to ternary hash codes by continuation
title_sort deep learning to ternary hash codes by continuation
publisher Wiley
publishDate 2021
url https://doaj.org/article/a5c841e2503147e9bb25a3892e2fa9f6
work_keys_str_mv AT mingruichen deeplearningtoternaryhashcodesbycontinuation
AT weiyuli deeplearningtoternaryhashcodesbycontinuation
AT weizhilu deeplearningtoternaryhashcodesbycontinuation
_version_ 1718420392730689536