Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy
Identifieur interne : 001547 ( Istex/Corpus ); précédent : 001546; suivant : 001548Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy
Auteurs : Thomas Udelhoven ; Brigitta SchüttSource :
- Chemometrics and Intelligent Laboratory Systems [ 0169-7439 ] ; 2000.
English descriptors
Abstract
Diffuse reflectance spectroscopy (0.4–2.5 μm) is evaluated as fast and non-destructive method for the analysis of sediments, characterised by a wide range of mineral constituents. Combined with feed-forward artificial neural networks (ANNs) this technique is used to estimate quantitatively the chemical composition from the sediments based on a supervised training with one model. The examined characteristics include contents of inorganic carbon, Fe, S, Al, Si, Ca, K, Mg and calcite. The efficiency of several learning algorithms (Backpropagation, Quickprop, Resilient propagation (Rprop), Cascade Correlation (CC)) is investigated. All learning algorithms perform well using principal component (PC) scores of the first derivative spectra as input for the supervised training. ANNs trained with Quickprop and Rprop produced most accurate estimations of the chemical characteristics and the performance was better than for standard multivariate statistical tools (stepwise multiple linear regression (SMLR), principal component analysis (PCA)). An interpretation of the results is given by a detailed consideration of the correlation structure among the chemical constituents.
Url:
DOI: 10.1016/S0169-7439(99)00069-6
Links to Exploration step
ISTEX:8C91855D112C248343F43E793E27E08A33B3A5D4Le document en format XML
<record><TEI wicri:istexFullTextTei="biblStruct"><teiHeader><fileDesc><titleStmt><title xml:lang="en">Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy</title><author><name sortKey="Udelhoven, Thomas" sort="Udelhoven, Thomas" uniqKey="Udelhoven T" first="Thomas" last="Udelhoven">Thomas Udelhoven</name><affiliation><mods:affiliation>E-mail: udelhove@uni-trier.de</mods:affiliation></affiliation><affiliation><mods:affiliation>Department of Remote Sensing, University of Trier, D-54286 TrierGermany</mods:affiliation></affiliation></author><author><name sortKey="Schutt, Brigitta" sort="Schutt, Brigitta" uniqKey="Schutt B" first="Brigitta" last="Schütt">Brigitta Schütt</name><affiliation><mods:affiliation>Department of Physical Geography, University of Trier, D-54286 TrierGermany</mods:affiliation></affiliation></author></titleStmt><publicationStmt><idno type="wicri:source">ISTEX</idno><idno type="RBID">ISTEX:8C91855D112C248343F43E793E27E08A33B3A5D4</idno><date when="2000" year="2000">2000</date><idno type="doi">10.1016/S0169-7439(99)00069-6</idno><idno type="url">https://api.istex.fr/document/8C91855D112C248343F43E793E27E08A33B3A5D4/fulltext/pdf</idno><idno type="wicri:Area/Istex/Corpus">001547</idno><idno type="wicri:explorRef" wicri:stream="Istex" wicri:step="Corpus" wicri:corpus="ISTEX">001547</idno></publicationStmt><sourceDesc><biblStruct><analytic><title level="a" type="main" xml:lang="en">Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy</title><author><name sortKey="Udelhoven, Thomas" sort="Udelhoven, Thomas" uniqKey="Udelhoven T" first="Thomas" last="Udelhoven">Thomas Udelhoven</name><affiliation><mods:affiliation>E-mail: udelhove@uni-trier.de</mods:affiliation></affiliation><affiliation><mods:affiliation>Department of Remote Sensing, University of Trier, D-54286 TrierGermany</mods:affiliation></affiliation></author><author><name sortKey="Schutt, Brigitta" sort="Schutt, Brigitta" uniqKey="Schutt B" first="Brigitta" last="Schütt">Brigitta Schütt</name><affiliation><mods:affiliation>Department of Physical Geography, University of Trier, D-54286 TrierGermany</mods:affiliation></affiliation></author></analytic><monogr></monogr><series><title level="j">Chemometrics and Intelligent Laboratory Systems</title><title level="j" type="abbrev">CHEMOM</title><idno type="ISSN">0169-7439</idno><imprint><publisher>ELSEVIER</publisher><date type="published" when="2000">2000</date><biblScope unit="volume">51</biblScope><biblScope unit="issue">1</biblScope><biblScope unit="page" from="9">9</biblScope><biblScope unit="page" to="22">22</biblScope></imprint><idno type="ISSN">0169-7439</idno></series><idno type="istex">8C91855D112C248343F43E793E27E08A33B3A5D4</idno><idno type="DOI">10.1016/S0169-7439(99)00069-6</idno><idno type="PII">S0169-7439(99)00069-6</idno></biblStruct></sourceDesc><seriesStmt><idno type="ISSN">0169-7439</idno></seriesStmt></fileDesc><profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Diffuse reflectance spectroscopy</term><term>Multivariate calibration</term><term>Neural networks</term></keywords></textClass><langUsage><language ident="en">en</language></langUsage></profileDesc></teiHeader><front><div type="abstract" xml:lang="en">Diffuse reflectance spectroscopy (0.4–2.5 μm) is evaluated as fast and non-destructive method for the analysis of sediments, characterised by a wide range of mineral constituents. Combined with feed-forward artificial neural networks (ANNs) this technique is used to estimate quantitatively the chemical composition from the sediments based on a supervised training with one model. The examined characteristics include contents of inorganic carbon, Fe, S, Al, Si, Ca, K, Mg and calcite. The efficiency of several learning algorithms (Backpropagation, Quickprop, Resilient propagation (Rprop), Cascade Correlation (CC)) is investigated. All learning algorithms perform well using principal component (PC) scores of the first derivative spectra as input for the supervised training. ANNs trained with Quickprop and Rprop produced most accurate estimations of the chemical characteristics and the performance was better than for standard multivariate statistical tools (stepwise multiple linear regression (SMLR), principal component analysis (PCA)). An interpretation of the results is given by a detailed consideration of the correlation structure among the chemical constituents.</div></front></TEI><istex><corpusName>elsevier</corpusName><author><json:item><name>Thomas Udelhoven</name><affiliations><json:string>E-mail: udelhove@uni-trier.de</json:string><json:string>Department of Remote Sensing, University of Trier, D-54286 TrierGermany</json:string></affiliations></json:item><json:item><name>Brigitta Schütt</name><affiliations><json:string>Department of Physical Geography, University of Trier, D-54286 TrierGermany</json:string></affiliations></json:item></author><subject><json:item><lang><json:string>eng</json:string></lang><value>Neural networks</value></json:item><json:item><lang><json:string>eng</json:string></lang><value>Diffuse reflectance spectroscopy</value></json:item><json:item><lang><json:string>eng</json:string></lang><value>Multivariate calibration</value></json:item></subject><language><json:string>eng</json:string></language><originalGenre><json:string>Full-length article</json:string></originalGenre><abstract>Diffuse reflectance spectroscopy (0.4–2.5 μm) is evaluated as fast and non-destructive method for the analysis of sediments, characterised by a wide range of mineral constituents. Combined with feed-forward artificial neural networks (ANNs) this technique is used to estimate quantitatively the chemical composition from the sediments based on a supervised training with one model. The examined characteristics include contents of inorganic carbon, Fe, S, Al, Si, Ca, K, Mg and calcite. The efficiency of several learning algorithms (Backpropagation, Quickprop, Resilient propagation (Rprop), Cascade Correlation (CC)) is investigated. All learning algorithms perform well using principal component (PC) scores of the first derivative spectra as input for the supervised training. ANNs trained with Quickprop and Rprop produced most accurate estimations of the chemical characteristics and the performance was better than for standard multivariate statistical tools (stepwise multiple linear regression (SMLR), principal component analysis (PCA)). An interpretation of the results is given by a detailed consideration of the correlation structure among the chemical constituents.</abstract><qualityIndicators><score>6.92</score><pdfVersion>1.2</pdfVersion><pdfPageSize>540 x 712 pts</pdfPageSize><refBibsNative>true</refBibsNative><keywordCount>3</keywordCount><abstractCharCount>1179</abstractCharCount><pdfWordCount>5711</pdfWordCount><pdfCharCount>36700</pdfCharCount><pdfPageCount>14</pdfPageCount><abstractWordCount>160</abstractWordCount></qualityIndicators><title>Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy</title><pii><json:string>S0169-7439(99)00069-6</json:string></pii><refBibs><json:item><author><json:item><name>G.R. Hunt</name></json:item><json:item><name>J.W. Salisbury</name></json:item></author><host><volume>1</volume><pages><last>300</last><first>283</first></pages><author></author><title>Modern Geology</title></host><title>Visible and near-infrared spectra of minerals and rocks: I Silicate minerals</title></json:item><json:item><author><json:item><name>G.R. Hunt</name></json:item><json:item><name>J.W. Salisbury</name></json:item><json:item><name>C.J. Lenhoff</name></json:item></author><host><volume>2</volume><pages><last>205</last><first>195</first></pages><author></author><title>Modern Geology</title></host><title>Visible and near-infrared spectra of minerals and rocks: III. Oxides and hydroxides</title></json:item><json:item><author><json:item><name>F. Despagne</name></json:item><json:item><name>D.L. Massart</name></json:item></author><host><volume>123</volume><pages><last>178</last><first>157</first></pages><author></author><title>Analyst</title></host><title>Neural networks in multivariate calibration</title></json:item><json:item><author><json:item><name>B.M. Wise</name></json:item><json:item><name>B.R. Holt</name></json:item><json:item><name>N.B. Gallagher</name></json:item><json:item><name>S. Lee</name></json:item></author><host><volume>30</volume><pages><last>89</last><first>81</first></pages><author></author><title>Chemometrics and Intelligent Laboratory Systems</title></host><title>A comparison of neural networks, non-linear biased regression and a genetic algorithm for dynamic model identification</title></json:item><json:item><author><json:item><name>M.S. Sánchez</name></json:item><json:item><name>L.A. Sarabia</name></json:item></author><host><volume>28</volume><pages><last>303</last><first>287</first></pages><author></author><title>Chemometrics and Intelligent Laboratory Systems</title></host><title>Efficiency of multi-layered feed-forward neural networks on classification in relation to linear discriminat analysis, quadratic discriminant analysis and regularized discriminant analysis</title></json:item><json:item><author><json:item><name>A. Panaye</name></json:item><json:item><name>J.-P. Doucet</name></json:item><json:item><name>B.T. Fan</name></json:item></author><host><volume>33</volume><pages><last>265</last><first>258</first></pages><author></author><title>Journal of Chemical Information and Computer Science</title></host><title>Topological approach of 13C NMR spectral simulation. Application to fuzzy substructures</title></json:item><json:item><author><json:item><name>D. Wienke</name></json:item><json:item><name>G. Kateman</name></json:item></author><host><volume>23</volume><pages><last>329</last><first>309</first></pages><author></author><title>Chemometrics and Intelligent Laboratory Systems</title></host><title>Adaptive resonance theory based artificial neural networks for treatment of opencategory problems in chemical pattern recognition: application to UV–VIS and IR spectroscopy</title></json:item><json:item><author><json:item><name>D.A. Cirovic</name></json:item></author><host><volume>16</volume><pages><last>155</last><first>148</first></pages><author></author><title>Trends in Analytical Chemistry</title></host><title>Feed-forward artificial neural networks: applications to spectroscopy</title></json:item><json:item><author><json:item><name>P. Zheng</name></json:item><json:item><name>P. Harrington</name></json:item><json:item><name>D.M. Davis</name></json:item></author><host><volume>33</volume><pages><last>132</last><first>121</first></pages><author></author><title>Chemometrics and Intelligent Laboratory Systems</title></host><title>Quantitative analysis of volatile organic compounds using ion mobility spectrometry and cascade correlation neural networks</title></json:item><json:item><author><json:item><name>W. Werther</name></json:item><json:item><name>H. Lohninger</name></json:item><json:item><name>F. Stanel</name></json:item><json:item><name>K. Varmuzza</name></json:item></author><host><volume>22</volume><pages><last>676</last><first>663</first></pages><author></author><title>Chemometrics and Intelligent Laboratory Systems</title></host><title>Classification of mass spectra. A comparison of yes/no classification methods for the recognition of simple structural properties</title></json:item><json:item><author><json:item><name>A. Eghbaldar</name></json:item><json:item><name>T.P. Forrest</name></json:item><json:item><name>D. Cabrol-Bass</name></json:item><json:item><name>A. Canmbon</name></json:item><json:item><name>J.-M. Guigonis</name></json:item></author><host><volume>36</volume><pages><last>643</last><first>637</first></pages><author></author><title>Journal of Chemical Information and Computer Science</title></host><title>Identification of structural features from mass spectrometry using a neural network approach: application to trimethylsilyl derivates used for medical diagnosis</title></json:item><json:item><host><author></author></host></json:item><json:item><author><json:item><name>E.W. Robb</name></json:item><json:item><name>M.E. Munk</name></json:item></author><host><volume>1</volume><pages><last>155</last><first>131</first></pages><author></author><title>Mikrochimica Acta</title></host><title>A neural network approach to infrared spectrum interpretation</title></json:item><json:item><author><json:item><name>C. Klawun</name></json:item><json:item><name>C.L. Wilkins</name></json:item></author><host><volume>36</volume><pages><last>257</last><first>249</first></pages><author></author><title>Journal of Chemical Information and Computer Science</title></host><title>Joint neural network interpretation of infrared and mass spectra</title></json:item><json:item><author><json:item><name>W. Wu</name></json:item><json:item><name>D.L. Massart</name></json:item></author><host><volume>35</volume><pages><last>135</last><first>127</first></pages><author></author><title>Chemometrics and Intelligent Laboratory Systems</title></host><title>Artificial neural networks in classification of NIR spectral data: selection of the input</title></json:item><json:item><author><json:item><name>W. Wu</name></json:item><json:item><name>B. Walczak</name></json:item><json:item><name>D.L. Massart</name></json:item><json:item><name>S. Heuerding</name></json:item><json:item><name>F. Erni</name></json:item><json:item><name>I.R. Last</name></json:item><json:item><name>K.A. Prebble</name></json:item></author><host><volume>33</volume><pages><last>46</last><first>35</first></pages><author></author><title>Chemometrics and Intelligent Laboratory Systems</title></host><title>Artificial neural networks in classification of NIR spectral data: design of the training set</title></json:item><json:item><author><json:item><name>H. Yang</name></json:item><json:item><name>P.R. Griffiths</name></json:item></author><host><volume>71</volume><pages><last>1161</last><first>751</first></pages><author></author><title>Analytical Chemistry</title></host><title>Application of multilayer feed-forward neural networks to automated compound identification in low-resolution open path FT-IR spectrometry</title></json:item><json:item><author><json:item><name>C. Cleva</name></json:item><json:item><name>C. Cachet</name></json:item><json:item><name>D. Cabrol-Bass</name></json:item><json:item><name>T.P. Forrest</name></json:item></author><host><volume>348</volume><pages><last>265</last><first>255</first></pages><author></author><title>Analytica Chimica Acta</title></host><title>Advantages of a hierarchical systems of neural-networks for the interpretation of infrared spectra in structure determination</title></json:item><json:item><host><author></author></host></json:item><json:item><host><volume>36</volume><pages><last>1638</last><first>1627</first></pages><author></author><title>Analytical Chemistry</title></host></json:item><json:item><host><author></author></host></json:item><json:item><host><author></author></host></json:item><json:item><host><pages><first>279</first></pages><author></author><title>Neuronale Netze: Optimierung durch Lernen und Evolution</title></host></json:item><json:item><host><author></author></host></json:item><json:item><host><author></author></host></json:item><json:item><host><author></author></host></json:item><json:item><author><json:item><name>H. Martens</name></json:item><json:item><name>T. Naes</name></json:item></author><host><pages><first>420</first></pages><author></author></host><title>Multivariate Calibration</title></json:item><json:item><author><json:item><name>T. Masters</name></json:item></author><host><pages><first>493</first></pages><author></author></host><title>Practical neural network recipes in C++</title></json:item><json:item><author><json:item><name>R. Riedmiller</name></json:item></author><host><volume>16</volume><pages><last>278</last><first>265</first></pages><author></author><title>Computer Standards and Interfaces</title></host><title>Advanced supervised learning in multilayer perceptrons - from backpropagation to adaptive learning algorithms</title></json:item><json:item><author><json:item><name>A.J. Obukhov</name></json:item><json:item><name>D.S. Orlov</name></json:item></author><host><volume>2</volume><pages><last>184</last><first>174</first></pages><author></author><title>Soviet Soil Science</title></host><title>Spectral reflectivity of the major soil groups and possibility of using diffuse reflection in soil investigations</title></json:item><json:item><host><author></author></host></json:item><json:item><host><author></author></host></json:item><json:item><author><json:item><name>M.F. Baumgardner</name></json:item><json:item><name>L.F. Silvia</name></json:item><json:item><name>L.L. Biehl</name></json:item><json:item><name>E.R. Stoner</name></json:item></author><host><volume>38</volume><pages><last>44</last><first>1</first></pages><author></author><title>Advances in Agronomy</title></host><title>Reflectance properties of soils</title></json:item><json:item><author><json:item><name>E. Ben-Dor</name></json:item><json:item><name>A. Banin</name></json:item></author><host><volume>150</volume><pages><last>270</last><first>259</first></pages><author></author><title>Soil Science</title></host><title>Near infrared analysis (NIRA) as a method to simultaneously evaluate spectral featuresless constituents in soils</title></json:item><json:item><host><author></author></host></json:item><json:item><author><json:item><name>P.A. Meyers</name></json:item><json:item><name>R. Ishiwatari</name></json:item></author><host><pages><last>209</last><first>185</first></pages><author></author><title>Organic Geochemistry — Principles and Applications</title></host><title>The early diagenesis of organic matter in lacustrine sediments</title></json:item><json:item><author><json:item><name>B. Schütt</name></json:item></author><host><volume>20</volume><pages><last>234</last><first>217</first></pages><author></author><title>Journal of Paleolimnology</title></host><title>Reconstruction of Holocene Paleoenvironments in the Endorheic Basin of Laguna de Gallocanta, Central Spain by investigation of mineralogical and geochemical characters from lacustrine sediments</title></json:item></refBibs><genre><json:string>research-article</json:string></genre><host><volume>51</volume><pii><json:string>S0169-7439(00)X0038-X</json:string></pii><pages><last>22</last><first>9</first></pages><issn><json:string>0169-7439</json:string></issn><issue>1</issue><genre><json:string>journal</json:string></genre><language><json:string>unknown</json:string></language><title>Chemometrics and Intelligent Laboratory Systems</title><publicationDate>2000</publicationDate></host><publicationDate>2000</publicationDate><copyrightDate>2000</copyrightDate><doi><json:string>10.1016/S0169-7439(99)00069-6</json:string></doi><id>8C91855D112C248343F43E793E27E08A33B3A5D4</id><score>1.232305</score><fulltext><json:item><extension>pdf</extension><original>true</original><mimetype>application/pdf</mimetype><uri>https://api.istex.fr/document/8C91855D112C248343F43E793E27E08A33B3A5D4/fulltext/pdf</uri></json:item><json:item><extension>zip</extension><original>false</original><mimetype>application/zip</mimetype><uri>https://api.istex.fr/document/8C91855D112C248343F43E793E27E08A33B3A5D4/fulltext/zip</uri></json:item><istex:fulltextTEI uri="https://api.istex.fr/document/8C91855D112C248343F43E793E27E08A33B3A5D4/fulltext/tei"><teiHeader><fileDesc><titleStmt><title level="a" type="main" xml:lang="en">Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy</title></titleStmt><publicationStmt><authority>ISTEX</authority><publisher>ELSEVIER</publisher><availability><p>©2000 Elsevier Science B.V.</p></availability><date>2000</date></publicationStmt><notesStmt><note type="content">Fig. 1: Spectral reflectance and its first derivative (a) and X-ray powder diffraction pattern (Cu Kα-radiation) (b) for Miocene calcareous bedrock from the Lagune de Sancho Gomez watershed, La Manche area. Sample's mineral fabric is shown in the diffractograme, sample's chemical composition is derived from X-ray fluorescence analysis.</note><note type="content">Fig. 2: Plot of predicted vs. measured chemical characteristics from the bulk samples (cross-validation). The predicted values are based on an ANN trained with Rprop and three-point averages from first derivative spectra. The relative prediction error (RPE) is defined as RMSCV/mean.</note><note type="content">Fig. 3: Correlation spectra of chemical characteristics from bulk samples.</note><note type="content">Table 1: Character of the sample set</note><note type="content">Table 2: Comparison of ANNs trained with different learning algorithms, based on first derivative spectra (three point averages, 716 input units) and their first 20 principal components (PC scores, 20 input units) Sum squared approximation errors (SSEs) are related to the cross-validation. SSEs and training cycles are calculated as averages from five independent ANN calibrations. The number of hidden units for CC was restricted to 75 for training with PC scores and to 1750 for three point averaged derivative spectra. Where: η+: starting value for all Δij, η−: upper limit for Δij, η1+: specifies the factor by which the update-value Δij is to be decreased when minimising the net error, η2+: specifies the factor by which the update-value Δij is to be increased when minimising the net error, η1−: specifies the factor by which the update-value Δij is to be decreased when maximising the covariance, η2: specifies the factor by which the update-value Δij is to be increased when maximising the convariance, η: learning parameter,μ: maximum growth parameter, ν: weight decay term (nomenclature is taken from [26]).</note><note type="content">Table 3: Root mean square error of cross-validation (RMSCV) of measured and predicted sediment properties ANNs were trained with Rprop.</note><note type="content">Table 4: Matrix of linear correlation coefficients of the soil's and sediment's chemical properties in the whole pattern set All significant coefficients (α=0.01) larger or equal 0.6 are printed.</note></notesStmt><sourceDesc><biblStruct type="inbook"><analytic><title level="a" type="main" xml:lang="en">Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy</title><author xml:id="author-1"><persName><forename type="first">Thomas</forename><surname>Udelhoven</surname></persName><email>udelhove@uni-trier.de</email><note type="correspondence"><p>Corresponding author. Tel.: +49-651-4594; fax: +49-651-3815</p></note><affiliation>Department of Remote Sensing, University of Trier, D-54286 TrierGermany</affiliation></author><author xml:id="author-2"><persName><forename type="first">Brigitta</forename><surname>Schütt</surname></persName><affiliation>Department of Physical Geography, University of Trier, D-54286 TrierGermany</affiliation></author></analytic><monogr><title level="j">Chemometrics and Intelligent Laboratory Systems</title><title level="j" type="abbrev">CHEMOM</title><idno type="pISSN">0169-7439</idno><idno type="PII">S0169-7439(00)X0038-X</idno><imprint><publisher>ELSEVIER</publisher><date type="published" when="2000"></date><biblScope unit="volume">51</biblScope><biblScope unit="issue">1</biblScope><biblScope unit="page" from="9">9</biblScope><biblScope unit="page" to="22">22</biblScope></imprint></monogr><idno type="istex">8C91855D112C248343F43E793E27E08A33B3A5D4</idno><idno type="DOI">10.1016/S0169-7439(99)00069-6</idno><idno type="PII">S0169-7439(99)00069-6</idno></biblStruct></sourceDesc></fileDesc><profileDesc><creation><date>2000</date></creation><langUsage><language ident="en">en</language></langUsage><abstract xml:lang="en"><p>Diffuse reflectance spectroscopy (0.4–2.5 μm) is evaluated as fast and non-destructive method for the analysis of sediments, characterised by a wide range of mineral constituents. Combined with feed-forward artificial neural networks (ANNs) this technique is used to estimate quantitatively the chemical composition from the sediments based on a supervised training with one model. The examined characteristics include contents of inorganic carbon, Fe, S, Al, Si, Ca, K, Mg and calcite. The efficiency of several learning algorithms (Backpropagation, Quickprop, Resilient propagation (Rprop), Cascade Correlation (CC)) is investigated. All learning algorithms perform well using principal component (PC) scores of the first derivative spectra as input for the supervised training. ANNs trained with Quickprop and Rprop produced most accurate estimations of the chemical characteristics and the performance was better than for standard multivariate statistical tools (stepwise multiple linear regression (SMLR), principal component analysis (PCA)). An interpretation of the results is given by a detailed consideration of the correlation structure among the chemical constituents.</p></abstract><textClass xml:lang="en"><keywords scheme="keyword"><list><head>Keywords</head><item><term>Neural networks</term></item><item><term>Diffuse reflectance spectroscopy</term></item><item><term>Multivariate calibration</term></item></list></keywords></textClass></profileDesc><revisionDesc><change when="2000">Published</change></revisionDesc></teiHeader></istex:fulltextTEI><json:item><extension>txt</extension><original>false</original><mimetype>text/plain</mimetype><uri>https://api.istex.fr/document/8C91855D112C248343F43E793E27E08A33B3A5D4/fulltext/txt</uri></json:item></fulltext><metadata><istex:metadataXml wicri:clean="Elsevier, elements deleted: ce:floats; body; tail"><istex:xmlDeclaration>version="1.0" encoding="utf-8"</istex:xmlDeclaration><istex:docType PUBLIC="-//ES//DTD journal article DTD version 4.5.2//EN//XML" URI="art452.dtd" name="istex:docType"><istex:entity SYSTEM="gr1" NDATA="IMAGE" name="gr1"></istex:entity><istex:entity SYSTEM="gr2" NDATA="IMAGE" name="gr2"></istex:entity><istex:entity SYSTEM="gr3" NDATA="IMAGE" name="gr3"></istex:entity></istex:docType><istex:document><converted-article version="4.5.2" docsubtype="fla" xml:lang="en"><item-info><jid>CHEMOM</jid><aid>1134</aid><ce:pii>S0169-7439(99)00069-6</ce:pii><ce:doi>10.1016/S0169-7439(99)00069-6</ce:doi><ce:copyright type="full-transfer" year="2000">Elsevier Science B.V.</ce:copyright></item-info><head><ce:title>Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy</ce:title><ce:author-group><ce:author><ce:given-name>Thomas</ce:given-name><ce:surname>Udelhoven</ce:surname><ce:cross-ref refid="AFF1"><ce:sup>a</ce:sup></ce:cross-ref><ce:cross-ref refid="CORR1">*</ce:cross-ref><ce:e-address>udelhove@uni-trier.de</ce:e-address></ce:author><ce:author><ce:indexed-name>Schutt</ce:indexed-name><ce:given-name>Brigitta</ce:given-name><ce:surname>Schütt</ce:surname><ce:cross-ref refid="AFF2"><ce:sup>b</ce:sup></ce:cross-ref></ce:author><ce:affiliation id="AFF1"><ce:label>a</ce:label><ce:textfn>Department of Remote Sensing, University of Trier, D-54286 TrierGermany</ce:textfn></ce:affiliation><ce:affiliation id="AFF2"><ce:label>b</ce:label><ce:textfn>Department of Physical Geography, University of Trier, D-54286 TrierGermany</ce:textfn></ce:affiliation><ce:correspondence id="CORR1"><ce:label>*</ce:label><ce:text>Corresponding author. Tel.: +49-651-4594; fax: +49-651-3815</ce:text></ce:correspondence></ce:author-group><ce:date-received day="29" month="7" year="1999"></ce:date-received><ce:date-accepted day="15" month="11" year="1999"></ce:date-accepted><ce:abstract><ce:section-title>Abstract</ce:section-title><ce:abstract-sec><ce:simple-para>Diffuse reflectance spectroscopy (0.4–2.5 μm) is evaluated as fast and non-destructive method for the analysis of sediments, characterised by a wide range of mineral constituents. Combined with feed-forward artificial neural networks (ANNs) this technique is used to estimate quantitatively the chemical composition from the sediments based on a supervised training with one model. The examined characteristics include contents of inorganic carbon, Fe, S, Al, Si, Ca, K, Mg and calcite. The efficiency of several learning algorithms (Backpropagation, Quickprop, Resilient propagation (Rprop), Cascade Correlation (CC)) is investigated. All learning algorithms perform well using principal component (PC) scores of the first derivative spectra as input for the supervised training. ANNs trained with Quickprop and Rprop produced most accurate estimations of the chemical characteristics and the performance was better than for standard multivariate statistical tools (stepwise multiple linear regression (SMLR), principal component analysis (PCA)). An interpretation of the results is given by a detailed consideration of the correlation structure among the chemical constituents.</ce:simple-para></ce:abstract-sec></ce:abstract><ce:keywords class="keyword"><ce:section-title>Keywords</ce:section-title><ce:keyword><ce:text>Neural networks</ce:text></ce:keyword><ce:keyword><ce:text>Diffuse reflectance spectroscopy</ce:text></ce:keyword><ce:keyword><ce:text>Multivariate calibration</ce:text></ce:keyword></ce:keywords></head></converted-article></istex:document></istex:metadataXml><mods version="3.6"><titleInfo lang="en"><title>Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy</title></titleInfo><titleInfo type="alternative" lang="en" contentType="CDATA"><title>Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy</title></titleInfo><name type="personal"><namePart type="given">Thomas</namePart><namePart type="family">Udelhoven</namePart><affiliation>E-mail: udelhove@uni-trier.de</affiliation><affiliation>Department of Remote Sensing, University of Trier, D-54286 TrierGermany</affiliation><description>Corresponding author. Tel.: +49-651-4594; fax: +49-651-3815</description><role><roleTerm type="text">author</roleTerm></role></name><name type="personal"><namePart type="given">Brigitta</namePart><namePart type="family">Schütt</namePart><affiliation>Department of Physical Geography, University of Trier, D-54286 TrierGermany</affiliation><role><roleTerm type="text">author</roleTerm></role></name><typeOfResource>text</typeOfResource><genre type="research-article" displayLabel="Full-length article"></genre><originInfo><publisher>ELSEVIER</publisher><dateIssued encoding="w3cdtf">2000</dateIssued><copyrightDate encoding="w3cdtf">2000</copyrightDate></originInfo><language><languageTerm type="code" authority="iso639-2b">eng</languageTerm><languageTerm type="code" authority="rfc3066">en</languageTerm></language><physicalDescription><internetMediaType>text/html</internetMediaType></physicalDescription><abstract lang="en">Diffuse reflectance spectroscopy (0.4–2.5 μm) is evaluated as fast and non-destructive method for the analysis of sediments, characterised by a wide range of mineral constituents. Combined with feed-forward artificial neural networks (ANNs) this technique is used to estimate quantitatively the chemical composition from the sediments based on a supervised training with one model. The examined characteristics include contents of inorganic carbon, Fe, S, Al, Si, Ca, K, Mg and calcite. The efficiency of several learning algorithms (Backpropagation, Quickprop, Resilient propagation (Rprop), Cascade Correlation (CC)) is investigated. All learning algorithms perform well using principal component (PC) scores of the first derivative spectra as input for the supervised training. ANNs trained with Quickprop and Rprop produced most accurate estimations of the chemical characteristics and the performance was better than for standard multivariate statistical tools (stepwise multiple linear regression (SMLR), principal component analysis (PCA)). An interpretation of the results is given by a detailed consideration of the correlation structure among the chemical constituents.</abstract><note type="content">Fig. 1: Spectral reflectance and its first derivative (a) and X-ray powder diffraction pattern (Cu Kα-radiation) (b) for Miocene calcareous bedrock from the Lagune de Sancho Gomez watershed, La Manche area. Sample's mineral fabric is shown in the diffractograme, sample's chemical composition is derived from X-ray fluorescence analysis.</note><note type="content">Fig. 2: Plot of predicted vs. measured chemical characteristics from the bulk samples (cross-validation). The predicted values are based on an ANN trained with Rprop and three-point averages from first derivative spectra. The relative prediction error (RPE) is defined as RMSCV/mean.</note><note type="content">Fig. 3: Correlation spectra of chemical characteristics from bulk samples.</note><note type="content">Table 1: Character of the sample set</note><note type="content">Table 2: Comparison of ANNs trained with different learning algorithms, based on first derivative spectra (three point averages, 716 input units) and their first 20 principal components (PC scores, 20 input units) Sum squared approximation errors (SSEs) are related to the cross-validation. SSEs and training cycles are calculated as averages from five independent ANN calibrations. The number of hidden units for CC was restricted to 75 for training with PC scores and to 1750 for three point averaged derivative spectra. Where: η+: starting value for all Δij, η−: upper limit for Δij, η1+: specifies the factor by which the update-value Δij is to be decreased when minimising the net error, η2+: specifies the factor by which the update-value Δij is to be increased when minimising the net error, η1−: specifies the factor by which the update-value Δij is to be decreased when maximising the covariance, η2: specifies the factor by which the update-value Δij is to be increased when maximising the convariance, η: learning parameter,μ: maximum growth parameter, ν: weight decay term (nomenclature is taken from [26]).</note><note type="content">Table 3: Root mean square error of cross-validation (RMSCV) of measured and predicted sediment properties ANNs were trained with Rprop.</note><note type="content">Table 4: Matrix of linear correlation coefficients of the soil's and sediment's chemical properties in the whole pattern set All significant coefficients (α=0.01) larger or equal 0.6 are printed.</note><subject lang="en"><genre>Keywords</genre><topic>Neural networks</topic><topic>Diffuse reflectance spectroscopy</topic><topic>Multivariate calibration</topic></subject><relatedItem type="host"><titleInfo><title>Chemometrics and Intelligent Laboratory Systems</title></titleInfo><titleInfo type="abbreviated"><title>CHEMOM</title></titleInfo><genre type="journal">journal</genre><originInfo><dateIssued encoding="w3cdtf">20000508</dateIssued></originInfo><identifier type="ISSN">0169-7439</identifier><identifier type="PII">S0169-7439(00)X0038-X</identifier><part><date>20000508</date><detail type="volume"><number>51</number><caption>vol.</caption></detail><detail type="issue"><number>1</number><caption>no.</caption></detail><extent unit="issue pages"><start>1</start><end>142</end></extent><extent unit="pages"><start>9</start><end>22</end></extent></part></relatedItem><identifier type="istex">8C91855D112C248343F43E793E27E08A33B3A5D4</identifier><identifier type="DOI">10.1016/S0169-7439(99)00069-6</identifier><identifier type="PII">S0169-7439(99)00069-6</identifier><accessCondition type="use and reproduction" contentType="copyright">©2000 Elsevier Science B.V.</accessCondition><recordInfo><recordContentSource>ELSEVIER</recordContentSource><recordOrigin>Elsevier Science B.V., ©2000</recordOrigin></recordInfo></mods></metadata><serie></serie></istex></record>Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Rhénanie/explor/UnivTrevesV1/Data/Istex/Corpus
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 001547 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Istex/Corpus/biblio.hfd -nk 001547 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien
|wiki= Wicri/Rhénanie
|area= UnivTrevesV1
|flux= Istex
|étape= Corpus
|type= RBID
|clé= ISTEX:8C91855D112C248343F43E793E27E08A33B3A5D4
|texte= Capability of feed-forward neural networks for a chemical evaluation of sediments with diffuse reflectance spectroscopy
}}
| This area was generated with Dilib version V0.6.31. |  |