Distributed algorithms for unmixing hyperspectral data using nonnegative matrix factorization with sparsity constraints

Stefan Robila, Daniel Ricart

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

4 Citations (Scopus)

Abstract

The ability to examine and extract the sources of data from hyperspectral images has become more and more important as the amount of data collected increases. Recent research has yielded better and better algorithms for unmixing this data to provide more accuracy. One such algorithm is Nonnegative Matrix Factorization which aims to approximate the sources of the known end result. An issue with current approaches is they are designed to be run sequentially and can be very computationally expensive. In this paper, ways of improving the performance of Sparse Nonnegative Matrix Factorization algorithms are introduced by utilizing distributed computing over a cluster of computers. The goal was to find ways of maximizing the throughput of a known algorithm without worrying about the accuracy of the algorithm itself (as this is shown through separate investigation). This was accomplished by testing out technologies such as MPI, POSIX threads and the OpenMP library. The aim was to compare and contrast different methods and find out what might be the optimal solution to allow for large data sets.

Original languageEnglish
Title of host publication2013 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013 - Proceedings
Pages2156-2159
Number of pages4
DOIs
StatePublished - 1 Dec 2013
Event2013 33rd IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013 - Melbourne, VIC, Australia
Duration: 21 Jul 201326 Jul 2013

Publication series

NameInternational Geoscience and Remote Sensing Symposium (IGARSS)

Other

Other2013 33rd IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013
CountryAustralia
CityMelbourne, VIC
Period21/07/1326/07/13

Fingerprint

Factorization
Parallel algorithms
matrix
Distributed computer systems
Throughput
Testing

Keywords

  • Hyperspectral data
  • Nonnegative Matrix Factorization
  • distributed computing
  • multithreading

Cite this

Robila, S., & Ricart, D. (2013). Distributed algorithms for unmixing hyperspectral data using nonnegative matrix factorization with sparsity constraints. In 2013 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013 - Proceedings (pp. 2156-2159). [6723241] (International Geoscience and Remote Sensing Symposium (IGARSS)). https://doi.org/10.1109/IGARSS.2013.6723241
Robila, Stefan ; Ricart, Daniel. / Distributed algorithms for unmixing hyperspectral data using nonnegative matrix factorization with sparsity constraints. 2013 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013 - Proceedings. 2013. pp. 2156-2159 (International Geoscience and Remote Sensing Symposium (IGARSS)).
@inproceedings{6ee0e353b818457aaaf29a5ab002958c,
title = "Distributed algorithms for unmixing hyperspectral data using nonnegative matrix factorization with sparsity constraints",
abstract = "The ability to examine and extract the sources of data from hyperspectral images has become more and more important as the amount of data collected increases. Recent research has yielded better and better algorithms for unmixing this data to provide more accuracy. One such algorithm is Nonnegative Matrix Factorization which aims to approximate the sources of the known end result. An issue with current approaches is they are designed to be run sequentially and can be very computationally expensive. In this paper, ways of improving the performance of Sparse Nonnegative Matrix Factorization algorithms are introduced by utilizing distributed computing over a cluster of computers. The goal was to find ways of maximizing the throughput of a known algorithm without worrying about the accuracy of the algorithm itself (as this is shown through separate investigation). This was accomplished by testing out technologies such as MPI, POSIX threads and the OpenMP library. The aim was to compare and contrast different methods and find out what might be the optimal solution to allow for large data sets.",
keywords = "Hyperspectral data, Nonnegative Matrix Factorization, distributed computing, multithreading",
author = "Stefan Robila and Daniel Ricart",
year = "2013",
month = "12",
day = "1",
doi = "10.1109/IGARSS.2013.6723241",
language = "English",
isbn = "9781479911141",
series = "International Geoscience and Remote Sensing Symposium (IGARSS)",
pages = "2156--2159",
booktitle = "2013 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013 - Proceedings",

}

Robila, S & Ricart, D 2013, Distributed algorithms for unmixing hyperspectral data using nonnegative matrix factorization with sparsity constraints. in 2013 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013 - Proceedings., 6723241, International Geoscience and Remote Sensing Symposium (IGARSS), pp. 2156-2159, 2013 33rd IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013, Melbourne, VIC, Australia, 21/07/13. https://doi.org/10.1109/IGARSS.2013.6723241

Distributed algorithms for unmixing hyperspectral data using nonnegative matrix factorization with sparsity constraints. / Robila, Stefan; Ricart, Daniel.

2013 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013 - Proceedings. 2013. p. 2156-2159 6723241 (International Geoscience and Remote Sensing Symposium (IGARSS)).

Research output: Chapter in Book/Report/Conference proceedingConference contributionResearchpeer-review

TY - GEN

T1 - Distributed algorithms for unmixing hyperspectral data using nonnegative matrix factorization with sparsity constraints

AU - Robila, Stefan

AU - Ricart, Daniel

PY - 2013/12/1

Y1 - 2013/12/1

N2 - The ability to examine and extract the sources of data from hyperspectral images has become more and more important as the amount of data collected increases. Recent research has yielded better and better algorithms for unmixing this data to provide more accuracy. One such algorithm is Nonnegative Matrix Factorization which aims to approximate the sources of the known end result. An issue with current approaches is they are designed to be run sequentially and can be very computationally expensive. In this paper, ways of improving the performance of Sparse Nonnegative Matrix Factorization algorithms are introduced by utilizing distributed computing over a cluster of computers. The goal was to find ways of maximizing the throughput of a known algorithm without worrying about the accuracy of the algorithm itself (as this is shown through separate investigation). This was accomplished by testing out technologies such as MPI, POSIX threads and the OpenMP library. The aim was to compare and contrast different methods and find out what might be the optimal solution to allow for large data sets.

AB - The ability to examine and extract the sources of data from hyperspectral images has become more and more important as the amount of data collected increases. Recent research has yielded better and better algorithms for unmixing this data to provide more accuracy. One such algorithm is Nonnegative Matrix Factorization which aims to approximate the sources of the known end result. An issue with current approaches is they are designed to be run sequentially and can be very computationally expensive. In this paper, ways of improving the performance of Sparse Nonnegative Matrix Factorization algorithms are introduced by utilizing distributed computing over a cluster of computers. The goal was to find ways of maximizing the throughput of a known algorithm without worrying about the accuracy of the algorithm itself (as this is shown through separate investigation). This was accomplished by testing out technologies such as MPI, POSIX threads and the OpenMP library. The aim was to compare and contrast different methods and find out what might be the optimal solution to allow for large data sets.

KW - Hyperspectral data

KW - Nonnegative Matrix Factorization

KW - distributed computing

KW - multithreading

UR - http://www.scopus.com/inward/record.url?scp=84894287402&partnerID=8YFLogxK

U2 - 10.1109/IGARSS.2013.6723241

DO - 10.1109/IGARSS.2013.6723241

M3 - Conference contribution

SN - 9781479911141

T3 - International Geoscience and Remote Sensing Symposium (IGARSS)

SP - 2156

EP - 2159

BT - 2013 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013 - Proceedings

ER -

Robila S, Ricart D. Distributed algorithms for unmixing hyperspectral data using nonnegative matrix factorization with sparsity constraints. In 2013 IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2013 - Proceedings. 2013. p. 2156-2159. 6723241. (International Geoscience and Remote Sensing Symposium (IGARSS)). https://doi.org/10.1109/IGARSS.2013.6723241