Quaternion-based Deep Belief Networks fine-tuning

Detalhes bibliográficos
Autor(a) principal: Papa, Joao Paulo [UNESP]
Data de Publicação: 2017
Outros Autores: Rosa, Gustavo H. [UNESP], Pereira, Danillo R. [UNESP], Yang, Xin-She
Tipo de documento: Artigo
Idioma: eng
Título da fonte: Repositório Institucional da UNESP
Texto Completo: http://dx.doi.org/10.1016/j.asoc.2017.06.046
http://hdl.handle.net/11449/163439
Resumo: Deep learning techniques have been paramount in the last years, mainly due to their outstanding results in a number of applications. In this paper, we address the issue of fine-tuning parameters of Deep Belief Networks by means of meta-heuristics in which real-valued decision variables are described by quaternions. Such approaches essentially perform optimization in fitness landscapes that are mapped to a different representation based on hypercomplex numbers that may generate smoother surfaces. We therefore can map the optimization process onto a new space representation that is more suitable to learning parameters. Also, we proposed two approaches based on Harmony Search and quaternions that outperform the state-of-the-art results obtained so far in three public datasets for the reconstruction of binary images. (C) 2017 Elsevier B.V. All rights reserved.
id UNSP_5e3fdda7a6c2ee4e5b6541fbd9b02889
oai_identifier_str oai:repositorio.unesp.br:11449/163439
network_acronym_str UNSP
network_name_str Repositório Institucional da UNESP
repository_id_str 2946
spelling Quaternion-based Deep Belief Networks fine-tuningDeep Belief NetworksQuaternionHarmony SearchDeep learning techniques have been paramount in the last years, mainly due to their outstanding results in a number of applications. In this paper, we address the issue of fine-tuning parameters of Deep Belief Networks by means of meta-heuristics in which real-valued decision variables are described by quaternions. Such approaches essentially perform optimization in fitness landscapes that are mapped to a different representation based on hypercomplex numbers that may generate smoother surfaces. We therefore can map the optimization process onto a new space representation that is more suitable to learning parameters. Also, we proposed two approaches based on Harmony Search and quaternions that outperform the state-of-the-art results obtained so far in three public datasets for the reconstruction of binary images. (C) 2017 Elsevier B.V. All rights reserved.Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq)Sao Paulo State Univ, Dept Comp, Av Eng Luiz Edmundo Carrijo Coube 14-01, BR-17033360 Bauru, SP, BrazilMiddlesex Univ, Sch Sci & Technol, London NW4 4BT, EnglandSao Paulo State Univ, Dept Comp, Av Eng Luiz Edmundo Carrijo Coube 14-01, BR-17033360 Bauru, SP, BrazilFAPESP: 2014/12236-1FAPESP: 2014/16250-9FAPESP: 2015/25739-4CNPq: 470571/2013-6CNPq: 306166/2014-3Elsevier B.V.Universidade Estadual Paulista (Unesp)Middlesex UnivPapa, Joao Paulo [UNESP]Rosa, Gustavo H. [UNESP]Pereira, Danillo R. [UNESP]Yang, Xin-She2018-11-26T17:42:02Z2018-11-26T17:42:02Z2017-11-01info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/article328-335application/pdfhttp://dx.doi.org/10.1016/j.asoc.2017.06.046Applied Soft Computing. Amsterdam: Elsevier Science Bv, v. 60, p. 328-335, 2017.1568-4946http://hdl.handle.net/11449/16343910.1016/j.asoc.2017.06.046WOS:000414072200024WOS000414072200024.pdfWeb of Sciencereponame:Repositório Institucional da UNESPinstname:Universidade Estadual Paulista (UNESP)instacron:UNESPengApplied Soft Computing1,199info:eu-repo/semantics/openAccess2024-04-23T16:11:00Zoai:repositorio.unesp.br:11449/163439Repositório InstitucionalPUBhttp://repositorio.unesp.br/oai/requestopendoar:29462024-08-05T23:11:25.995219Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)false
dc.title.none.fl_str_mv Quaternion-based Deep Belief Networks fine-tuning
title Quaternion-based Deep Belief Networks fine-tuning
spellingShingle Quaternion-based Deep Belief Networks fine-tuning
Papa, Joao Paulo [UNESP]
Deep Belief Networks
Quaternion
Harmony Search
title_short Quaternion-based Deep Belief Networks fine-tuning
title_full Quaternion-based Deep Belief Networks fine-tuning
title_fullStr Quaternion-based Deep Belief Networks fine-tuning
title_full_unstemmed Quaternion-based Deep Belief Networks fine-tuning
title_sort Quaternion-based Deep Belief Networks fine-tuning
author Papa, Joao Paulo [UNESP]
author_facet Papa, Joao Paulo [UNESP]
Rosa, Gustavo H. [UNESP]
Pereira, Danillo R. [UNESP]
Yang, Xin-She
author_role author
author2 Rosa, Gustavo H. [UNESP]
Pereira, Danillo R. [UNESP]
Yang, Xin-She
author2_role author
author
author
dc.contributor.none.fl_str_mv Universidade Estadual Paulista (Unesp)
Middlesex Univ
dc.contributor.author.fl_str_mv Papa, Joao Paulo [UNESP]
Rosa, Gustavo H. [UNESP]
Pereira, Danillo R. [UNESP]
Yang, Xin-She
dc.subject.por.fl_str_mv Deep Belief Networks
Quaternion
Harmony Search
topic Deep Belief Networks
Quaternion
Harmony Search
description Deep learning techniques have been paramount in the last years, mainly due to their outstanding results in a number of applications. In this paper, we address the issue of fine-tuning parameters of Deep Belief Networks by means of meta-heuristics in which real-valued decision variables are described by quaternions. Such approaches essentially perform optimization in fitness landscapes that are mapped to a different representation based on hypercomplex numbers that may generate smoother surfaces. We therefore can map the optimization process onto a new space representation that is more suitable to learning parameters. Also, we proposed two approaches based on Harmony Search and quaternions that outperform the state-of-the-art results obtained so far in three public datasets for the reconstruction of binary images. (C) 2017 Elsevier B.V. All rights reserved.
publishDate 2017
dc.date.none.fl_str_mv 2017-11-01
2018-11-26T17:42:02Z
2018-11-26T17:42:02Z
dc.type.status.fl_str_mv info:eu-repo/semantics/publishedVersion
dc.type.driver.fl_str_mv info:eu-repo/semantics/article
format article
status_str publishedVersion
dc.identifier.uri.fl_str_mv http://dx.doi.org/10.1016/j.asoc.2017.06.046
Applied Soft Computing. Amsterdam: Elsevier Science Bv, v. 60, p. 328-335, 2017.
1568-4946
http://hdl.handle.net/11449/163439
10.1016/j.asoc.2017.06.046
WOS:000414072200024
WOS000414072200024.pdf
url http://dx.doi.org/10.1016/j.asoc.2017.06.046
http://hdl.handle.net/11449/163439
identifier_str_mv Applied Soft Computing. Amsterdam: Elsevier Science Bv, v. 60, p. 328-335, 2017.
1568-4946
10.1016/j.asoc.2017.06.046
WOS:000414072200024
WOS000414072200024.pdf
dc.language.iso.fl_str_mv eng
language eng
dc.relation.none.fl_str_mv Applied Soft Computing
1,199
dc.rights.driver.fl_str_mv info:eu-repo/semantics/openAccess
eu_rights_str_mv openAccess
dc.format.none.fl_str_mv 328-335
application/pdf
dc.publisher.none.fl_str_mv Elsevier B.V.
publisher.none.fl_str_mv Elsevier B.V.
dc.source.none.fl_str_mv Web of Science
reponame:Repositório Institucional da UNESP
instname:Universidade Estadual Paulista (UNESP)
instacron:UNESP
instname_str Universidade Estadual Paulista (UNESP)
instacron_str UNESP
institution UNESP
reponame_str Repositório Institucional da UNESP
collection Repositório Institucional da UNESP
repository.name.fl_str_mv Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)
repository.mail.fl_str_mv
_version_ 1808129497901301760