An optimized unsupervised manifold learning algorithm for manycore architectures

dc.contributor.authorBaldassin, Alexandro [UNESP]
dc.contributor.authorWeng, Ying
dc.contributor.authorGuimarães Pedronette, Daniel Carlos [UNESP]
dc.contributor.authorAlmeida, Jurandy
dc.contributor.institutionUniversidade Estadual Paulista (Unesp)
dc.contributor.institutionBangor University
dc.contributor.institutionUniversidade Federal de São Paulo (UNIFESP)
dc.date.accessioned2018-12-11T16:53:57Z
dc.date.available2018-12-11T16:53:57Z
dc.date.issued2018-01-01
dc.description.abstractMultimedia 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.en
dc.description.affiliationDepartment of Statistics Applied Mathematics and Computing São Paulo State University – UNESP
dc.description.affiliationSchool of Computer Science Bangor University
dc.description.affiliationInstituto de Ciência e Tecnologia Universidade Federal de São Paulo – UNIFESP
dc.description.affiliationUnespDepartment of Statistics Applied Mathematics and Computing São Paulo State University – UNESP
dc.identifierhttp://dx.doi.org/10.1016/j.ins.2018.06.023
dc.identifier.citationInformation Sciences.
dc.identifier.doi10.1016/j.ins.2018.06.023
dc.identifier.file2-s2.0-85048708990.pdf
dc.identifier.issn0020-0255
dc.identifier.scopus2-s2.0-85048708990
dc.identifier.urihttp://hdl.handle.net/11449/171114
dc.language.isoeng
dc.relation.ispartofInformation Sciences
dc.relation.ispartofsjr1,635
dc.rights.accessRightsAcesso aberto
dc.sourceScopus
dc.subjectEfficiency
dc.subjectMultimedia retrieval
dc.subjectParallelism
dc.subjectScalability
dc.subjectUnsupervised learning
dc.titleAn optimized unsupervised manifold learning algorithm for manycore architecturesen
dc.typeArtigo
unesp.author.lattes4738829911864396[1]
unesp.author.orcid0000-0001-8824-3055[1]
unesp.author.orcid0000-0002-4998-6996[4]

Arquivos

Pacote Original
Agora exibindo 1 - 1 de 1
Carregando...
Imagem de Miniatura
Nome:
2-s2.0-85048708990.pdf
Tamanho:
1.15 MB
Formato:
Adobe Portable Document Format
Descrição:

Coleções