Sequential and parallel feature extraction in hyperspectral data using nonnegative matrix factorization

Stefan Robila, Lukasz G. Maciak

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

7 Citations (Scopus)

Abstract

Feature extraction refers to the groups of techniques that, when applied to large dimensional and redundant data result in significant dimensionality reduction while preserving or even enhancing the information content. Among various techniques investigated for feature extraction, of new interest is Nonnegative Matrix Factorization (NMF). In NMF, it is assumed that the data is formed as a linear nonnegative combination of positive sources and the NMF solution recovers the original sources and the mixing matrix. In this paper, we first look at ways NMF can be applied for feature extraction in hyperspectral imagery a data known for large sizes and redundancy. While some of the associations are natural to linear mixing model (LMM - that assumes that hyperspectral images are formed as a linear mixture of endmember information), we also show NMF to be a slow method. To counter this, we investigate alternative solutions such as projected NMF approaches and provide an insight to how parallel implementations would contribute to speedup. Experimental results on various data show projected NMF outperforming regular NMF with parallel implementations providing a promising speedup advantage.

Original languageEnglish
Title of host publication2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT
Pages90-96
Number of pages7
DOIs
StatePublished - 1 Dec 2007
Event2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT - Farmingdale, NY, United States
Duration: 4 May 20074 May 2007

Other

Other2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT
CountryUnited States
CityFarmingdale, NY
Period4/05/074/05/07

Fingerprint

Factorization
Feature extraction
Redundancy

Keywords

  • Hyperspectral data
  • Linear algorithms
  • Linear unmixing
  • Nonnegative matrix factorization
  • Remote sensing

Cite this

Robila, S., & Maciak, L. G. (2007). Sequential and parallel feature extraction in hyperspectral data using nonnegative matrix factorization. In 2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT (pp. 90-96). [4312637] https://doi.org/10.1109/LISAT.2007.4312637
Robila, Stefan ; Maciak, Lukasz G. / Sequential and parallel feature extraction in hyperspectral data using nonnegative matrix factorization. 2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT. 2007. pp. 90-96
@inproceedings{93a950a2782545768934520757e6fe4a,
title = "Sequential and parallel feature extraction in hyperspectral data using nonnegative matrix factorization",
abstract = "Feature extraction refers to the groups of techniques that, when applied to large dimensional and redundant data result in significant dimensionality reduction while preserving or even enhancing the information content. Among various techniques investigated for feature extraction, of new interest is Nonnegative Matrix Factorization (NMF). In NMF, it is assumed that the data is formed as a linear nonnegative combination of positive sources and the NMF solution recovers the original sources and the mixing matrix. In this paper, we first look at ways NMF can be applied for feature extraction in hyperspectral imagery a data known for large sizes and redundancy. While some of the associations are natural to linear mixing model (LMM - that assumes that hyperspectral images are formed as a linear mixture of endmember information), we also show NMF to be a slow method. To counter this, we investigate alternative solutions such as projected NMF approaches and provide an insight to how parallel implementations would contribute to speedup. Experimental results on various data show projected NMF outperforming regular NMF with parallel implementations providing a promising speedup advantage.",
keywords = "Hyperspectral data, Linear algorithms, Linear unmixing, Nonnegative matrix factorization, Remote sensing",
author = "Stefan Robila and Maciak, {Lukasz G.}",
year = "2007",
month = "12",
day = "1",
doi = "10.1109/LISAT.2007.4312637",
language = "English",
isbn = "1424413028",
pages = "90--96",
booktitle = "2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT",

}

Robila, S & Maciak, LG 2007, Sequential and parallel feature extraction in hyperspectral data using nonnegative matrix factorization. in 2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT., 4312637, pp. 90-96, 2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT, Farmingdale, NY, United States, 4/05/07. https://doi.org/10.1109/LISAT.2007.4312637

Sequential and parallel feature extraction in hyperspectral data using nonnegative matrix factorization. / Robila, Stefan; Maciak, Lukasz G.

2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT. 2007. p. 90-96 4312637.

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

TY - GEN

T1 - Sequential and parallel feature extraction in hyperspectral data using nonnegative matrix factorization

AU - Robila, Stefan

AU - Maciak, Lukasz G.

PY - 2007/12/1

Y1 - 2007/12/1

N2 - Feature extraction refers to the groups of techniques that, when applied to large dimensional and redundant data result in significant dimensionality reduction while preserving or even enhancing the information content. Among various techniques investigated for feature extraction, of new interest is Nonnegative Matrix Factorization (NMF). In NMF, it is assumed that the data is formed as a linear nonnegative combination of positive sources and the NMF solution recovers the original sources and the mixing matrix. In this paper, we first look at ways NMF can be applied for feature extraction in hyperspectral imagery a data known for large sizes and redundancy. While some of the associations are natural to linear mixing model (LMM - that assumes that hyperspectral images are formed as a linear mixture of endmember information), we also show NMF to be a slow method. To counter this, we investigate alternative solutions such as projected NMF approaches and provide an insight to how parallel implementations would contribute to speedup. Experimental results on various data show projected NMF outperforming regular NMF with parallel implementations providing a promising speedup advantage.

AB - Feature extraction refers to the groups of techniques that, when applied to large dimensional and redundant data result in significant dimensionality reduction while preserving or even enhancing the information content. Among various techniques investigated for feature extraction, of new interest is Nonnegative Matrix Factorization (NMF). In NMF, it is assumed that the data is formed as a linear nonnegative combination of positive sources and the NMF solution recovers the original sources and the mixing matrix. In this paper, we first look at ways NMF can be applied for feature extraction in hyperspectral imagery a data known for large sizes and redundancy. While some of the associations are natural to linear mixing model (LMM - that assumes that hyperspectral images are formed as a linear mixture of endmember information), we also show NMF to be a slow method. To counter this, we investigate alternative solutions such as projected NMF approaches and provide an insight to how parallel implementations would contribute to speedup. Experimental results on various data show projected NMF outperforming regular NMF with parallel implementations providing a promising speedup advantage.

KW - Hyperspectral data

KW - Linear algorithms

KW - Linear unmixing

KW - Nonnegative matrix factorization

KW - Remote sensing

UR - http://www.scopus.com/inward/record.url?scp=47949130385&partnerID=8YFLogxK

U2 - 10.1109/LISAT.2007.4312637

DO - 10.1109/LISAT.2007.4312637

M3 - Conference contribution

SN - 1424413028

SN - 9781424413027

SP - 90

EP - 96

BT - 2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT

ER -

Robila S, Maciak LG. Sequential and parallel feature extraction in hyperspectral data using nonnegative matrix factorization. In 2007 IEEE Long Island Systems, Applications and Technology Conference, LISAT. 2007. p. 90-96. 4312637 https://doi.org/10.1109/LISAT.2007.4312637