An optimized unsupervised manifold learning algorithm for manycore architectures
Autor(a) principal: | |
---|---|
Data de Publicação: | 2018 |
Outros Autores: | , , |
Tipo de documento: | Artigo |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UNESP |
Texto Completo: | http://dx.doi.org/10.1016/j.ins.2018.06.023 http://hdl.handle.net/11449/171114 |
Resumo: | Multimedia data, such as images and videos, has become very popular in people's daily life as a result of the widespread use of mobile devices. The ever-increasing amount of such data, along with the necessity for real-time retrieval, has lead to the development of new methods that can process them in a timely fashion with acceptable accuracy. In this paper, we study the performance of ReckNN, an unsupervised manifold learning algorithm based on the reciprocal neighbourhood and the authority of ranked lists. Most of the related work in this field do not fully investigate optimization strategies, an aspect that is becoming more important with the high availability of manycore machines. In order to address that issue, we fully investigate optimization opportunities in this article and make the following three main contributions. Firstly, we develop an efficient and scalable method for storing and accessing the distances between objects (e.g., video or image) based on dictionaries. Secondly, we employ memoization to speed up the computation of authority scores, leading to a significant performance gain even on single-core architectures. Lastly, we devise and implement several parallelization strategies and show that they are scalable on a 72-core Intel machine. The experimental results with MPEG-7, Corel5k and MediaEval benchmarks show that the optimized ReckNN delivers both efficiency and scalability, highlighting the importance of the proposed optimizations for manycore machines. |
id |
UNSP_693ddf8c44591c4dcd03610d7b9dfa64 |
---|---|
oai_identifier_str |
oai:repositorio.unesp.br:11449/171114 |
network_acronym_str |
UNSP |
network_name_str |
Repositório Institucional da UNESP |
repository_id_str |
2946 |
spelling |
An optimized unsupervised manifold learning algorithm for manycore architecturesEfficiencyMultimedia retrievalParallelismScalabilityUnsupervised learningMultimedia data, such as images and videos, has become very popular in people's daily life as a result of the widespread use of mobile devices. The ever-increasing amount of such data, along with the necessity for real-time retrieval, has lead to the development of new methods that can process them in a timely fashion with acceptable accuracy. In this paper, we study the performance of ReckNN, an unsupervised manifold learning algorithm based on the reciprocal neighbourhood and the authority of ranked lists. Most of the related work in this field do not fully investigate optimization strategies, an aspect that is becoming more important with the high availability of manycore machines. In order to address that issue, we fully investigate optimization opportunities in this article and make the following three main contributions. Firstly, we develop an efficient and scalable method for storing and accessing the distances between objects (e.g., video or image) based on dictionaries. Secondly, we employ memoization to speed up the computation of authority scores, leading to a significant performance gain even on single-core architectures. Lastly, we devise and implement several parallelization strategies and show that they are scalable on a 72-core Intel machine. The experimental results with MPEG-7, Corel5k and MediaEval benchmarks show that the optimized ReckNN delivers both efficiency and scalability, highlighting the importance of the proposed optimizations for manycore machines.Department of Statistics Applied Mathematics and Computing São Paulo State University – UNESPSchool of Computer Science Bangor UniversityInstituto de Ciência e Tecnologia Universidade Federal de São Paulo – UNIFESPDepartment of Statistics Applied Mathematics and Computing São Paulo State University – UNESPUniversidade Estadual Paulista (Unesp)Bangor UniversityUniversidade Federal de São Paulo (UNIFESP)Baldassin, Alexandro [UNESP]Weng, YingGuimarães Pedronette, Daniel Carlos [UNESP]Almeida, Jurandy2018-12-11T16:53:57Z2018-12-11T16:53:57Z2018-01-01info:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/articleapplication/pdfhttp://dx.doi.org/10.1016/j.ins.2018.06.023Information Sciences.0020-0255http://hdl.handle.net/11449/17111410.1016/j.ins.2018.06.0232-s2.0-850487089902-s2.0-85048708990.pdfScopusreponame:Repositório Institucional da UNESPinstname:Universidade Estadual Paulista (UNESP)instacron:UNESPengInformation Sciences1,635info:eu-repo/semantics/openAccess2023-10-23T06:11:12Zoai:repositorio.unesp.br:11449/171114Repositório InstitucionalPUBhttp://repositorio.unesp.br/oai/requestopendoar:29462024-08-05T15:46:11.958965Repositório Institucional da UNESP - Universidade Estadual Paulista (UNESP)false |
dc.title.none.fl_str_mv |
An optimized unsupervised manifold learning algorithm for manycore architectures |
title |
An optimized unsupervised manifold learning algorithm for manycore architectures |
spellingShingle |
An optimized unsupervised manifold learning algorithm for manycore architectures Baldassin, Alexandro [UNESP] Efficiency Multimedia retrieval Parallelism Scalability Unsupervised learning |
title_short |
An optimized unsupervised manifold learning algorithm for manycore architectures |
title_full |
An optimized unsupervised manifold learning algorithm for manycore architectures |
title_fullStr |
An optimized unsupervised manifold learning algorithm for manycore architectures |
title_full_unstemmed |
An optimized unsupervised manifold learning algorithm for manycore architectures |
title_sort |
An optimized unsupervised manifold learning algorithm for manycore architectures |
author |
Baldassin, Alexandro [UNESP] |
author_facet |
Baldassin, Alexandro [UNESP] Weng, Ying Guimarães Pedronette, Daniel Carlos [UNESP] Almeida, Jurandy |
author_role |
author |
author2 |
Weng, Ying Guimarães Pedronette, Daniel Carlos [UNESP] Almeida, Jurandy |
author2_role |
author author author |
dc.contributor.none.fl_str_mv |
Universidade Estadual Paulista (Unesp) Bangor University Universidade Federal de São Paulo (UNIFESP) |
dc.contributor.author.fl_str_mv |
Baldassin, Alexandro [UNESP] Weng, Ying Guimarães Pedronette, Daniel Carlos [UNESP] Almeida, Jurandy |
dc.subject.por.fl_str_mv |
Efficiency Multimedia retrieval Parallelism Scalability Unsupervised learning |
topic |
Efficiency Multimedia retrieval Parallelism Scalability Unsupervised learning |
description |
Multimedia data, such as images and videos, has become very popular in people's daily life as a result of the widespread use of mobile devices. The ever-increasing amount of such data, along with the necessity for real-time retrieval, has lead to the development of new methods that can process them in a timely fashion with acceptable accuracy. In this paper, we study the performance of ReckNN, an unsupervised manifold learning algorithm based on the reciprocal neighbourhood and the authority of ranked lists. Most of the related work in this field do not fully investigate optimization strategies, an aspect that is becoming more important with the high availability of manycore machines. In order to address that issue, we fully investigate optimization opportunities in this article and make the following three main contributions. Firstly, we develop an efficient and scalable method for storing and accessing the distances between objects (e.g., video or image) based on dictionaries. Secondly, we employ memoization to speed up the computation of authority scores, leading to a significant performance gain even on single-core architectures. Lastly, we devise and implement several parallelization strategies and show that they are scalable on a 72-core Intel machine. The experimental results with MPEG-7, Corel5k and MediaEval benchmarks show that the optimized ReckNN delivers both efficiency and scalability, highlighting the importance of the proposed optimizations for manycore machines. |
publishDate |
2018 |
dc.date.none.fl_str_mv |
2018-12-11T16:53:57Z 2018-12-11T16:53:57Z 2018-01-01 |
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.ins.2018.06.023 Information Sciences. 0020-0255 http://hdl.handle.net/11449/171114 10.1016/j.ins.2018.06.023 2-s2.0-85048708990 2-s2.0-85048708990.pdf |
url |
http://dx.doi.org/10.1016/j.ins.2018.06.023 http://hdl.handle.net/11449/171114 |
identifier_str_mv |
Information Sciences. 0020-0255 10.1016/j.ins.2018.06.023 2-s2.0-85048708990 2-s2.0-85048708990.pdf |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.none.fl_str_mv |
Information Sciences 1,635 |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
application/pdf |
dc.source.none.fl_str_mv |
Scopus 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_ |
1808128559632351232 |