Data fusion methodology and applications ebook pdf

Page 1

Data Fusion Methodology and Applications

- eBook PDF

Visit to download the full and correct content document: https://ebooksecure.com/download/data-fusion-methodology-and-applications-ebookpdf/

DATAFUSIONMETHODOLOGY ANDAPPLICATIONS
DATAHANDLINGINSCIENCEANDTECHNOLOGY Volume31

Thispageintentionallyleftblank

DATAFUSION METHODOLOGYAND APPLICATIONS

BEATAWALCZAK

LUTGARDEBUYDENS

DATAHANDLINGINSCIENCE ANDTECHNOLOGY Volume31

Elsevier

Radarweg29,POBox211,1000AEAmsterdam,Netherlands

TheBoulevard,LangfordLane,Kidlington,OxfordOX51GB,UnitedKingdom 50HampshireStreet,5thFloor,Cambridge,MA02139,UnitedStates

Copyright © 2019ElsevierB.V.Allrightsreserved.

Nopartofthispublicationmaybereproducedortransmittedinanyformorbyanymeans, electronicormechanical,includingphotocopying,recording,oranyinformationstorageand retrievalsystem,withoutpermissioninwritingfromthePublisher.Detailsonhowtoseek permission,furtherinformationaboutthePublisher’spermissionspoliciesandourarrangements withorganizationssuchastheCopyrightClearanceCenterandtheCopyrightLicensingAgency,can befoundatourwebsite: www.elsevier.com/permissions

Thisbookandtheindividualcontributionscontainedinitareprotectedundercopyrightbythe Publisher(otherthanasmaybenotedherein).

Notices

Knowledgeandbestpracticeinthis fieldareconstantlychanging.Asnewresearchandexperience broadenourunderstanding,changesinresearchmethods,professionalpractices,ormedical treatmentmaybecomenecessary.

Practitionersandresearchersmustalwaysrelyontheirownexperienceandknowledgein evaluatingandusinganyinformation,methods,compounds,orexperimentsdescribedherein.In usingsuchinformationormethodstheyshouldbemindfuloftheirownsafetyandthesafetyof others,includingpartiesforwhomtheyhaveaprofessionalresponsibility.

Tothefullestextentofthelaw,neitherAACCInorthePublisher,northeauthors,contributors,or editors,assumeanyliabilityforanyinjuryand/ordamagetopersonsorpropertyasamatterof productsliability,negligenceorotherwise,orfromanyuseoroperationofanymethods, products,instructions,orideascontainedinthematerialherein.

LibraryofCongressCataloging-in-PublicationData

AcatalogrecordforthisbookisavailablefromtheLibraryofCongress

BritishLibraryCataloguing-in-PublicationData

AcataloguerecordforthisbookisavailablefromtheBritishLibrary

ISBN:978-0-444-63984-4

ISSN:0922-3487

ForinformationonallElsevierpublicationsvisitourwebsiteat https://www.elsevier.com/books-and-journals

Publisher: SusanDennis

AcquisitionEditor: KathrynMorrissey

EditorialProjectManager: LauraOkidi

ProductionProjectManager: PremKumarKaliamoorthi

CoverDesigner: ChristianBilbow

CoverDrawing: AnitaAccorsi

TypesetbyTNQTechnologies

Contributorsix Prefacexi

1.Introduction:WaysandMeanstoDealWithDataFrom MultipleSources

MARINACOCCHI

1.Motivation1

2.AFrameworkforLow-LevelDataFusion

AGEK.SMILDEANDIVENVANMECHELEN

1.IntroductionandMotivation27

3.GeneralFramingofLow-,Mid-,andHigh-LevelDataFusion WithExamplesintheLifeSciences

AGNIESZKASMOLINSKA,JASPERENGEL,EWASZYMANSKA,LUTGARDEBUYDENS,AND LIONELBLANCHET

1.Introduction51

2.DataSampling,Measurements,andPreprocessing54

3.DataFusionStrategy55

4.DataFusionStrategieswithExamples65

5.InterpretationoftheOutcomes72

6.Conclusions75 References76

Contents
References22
2.Context,Definition2 3.MainApproaches6 4.RemarksintheUser’sPerspective17
2.DataStructures31 3.FrameworkforLow-LevelDataFusion32 4.CommonandDistinctComponents38 5.Examples41 6.Conclusions47 References47
v

4.NumericalOptimization-BasedAlgorithmsforDataFusion

N.VERVLIETANDL.DELATHAUWER

1.Introduction81

2.NumericalOptimizationforTensorDecompositions85

3.CanonicalPolyadicDecomposition91

4.ConstrainedDecompositions100

5.CoupledDecompositions111

6.Large-ScaleComputations117 References122

5.RecentAdvancesinHigh-LevelFusionMethodstoClassify MultipleAnalyticalChemicalData

D.BALLABIO,R.TODESCHINI,ANDV.CONSONNI

1.Introduction129

6.TheSequentialandOrthogonalizedPLSRegressionfor MultiblockRegression:Theory,Examples,andExtensions

ALESSANDRABIANCOLILLOANDTORMODNÆS

1.Introduction157

2.HowItStarted158

3.ModelandAlgorithm158

4.SomeMathematicalFormulaeandProperties160

5.HowtoChoosetheOptimalNumberofComponents161

6.HowtoInterprettheModels162

7.SomeFurtherPropertiesoftheSO-PLSMethod163

8.ExamplesofStandardSO-PLSRegression165

9.ExtensionsandModificationsofSO-PLS167

10.Conclusions175 References176

7.ComDimMethodsfortheAnalysisofMultiblock DatainaDataFusionPerspective V.CARIOU,D.JOUAN-RIMBAUDBOUVERESSE,E.M.QANNARI,ANDD.N.RUTLEDGE

1.Introduction179

2.ComDimAnalysis181

3.P-ComDimAnalysis185

4.Path-ComDimAnalysis189

5.Software191

References153
2.Methods132 3.ApplicationonAnalyticalData144 4.Results148 5.Conclusions152
CONTENTS vi

6.Illustration191

7.Conclusion202 References202

8.DataFusionbyMultivariateCurveResolution

ANNADEJUANANDR.TAULER

1.Introduction.GeneralMultivariateCurveResolutionFramework. WhytoUseItinDataFusion?205

2.DataFusionStructuresinMCR.MultisetAnalysis208

3.ConstraintsinMCR.VersatilityLinkedtoDataFusion. HybridModels(Hard Soft,Bilinear/Multilinear)211

4.LimitationsOvercomebyMultisetMCRAnalysis.BreakingRankDeficiency andDecreasingAmbiguity218

5.AdditionalOutcomesofMCRMultisetAnalysis.TheHiddenDimensions221

6.DataFusionFields222

7.Conclusions227 References228

9.DealingWithDataHeterogeneityinaDataFusionPerspective:

Models,Methodologies,andAlgorithms

FEDERICAMANDREOLIANDMANUELAMONTANGERO

1.Introduction235

2.OverviewofLifeScienceDataSources237

3.AddressingDataHeterogeneity239

4.LatestTrendsandChallenges255

5.Conclusions264 References265

10.DataFusionStrategiesinFoodAnalysis

ALESSANDRABIANCOLILLO,RICARDBOQUE ´ ,MARINACOCCHI,ANDFEDERICOMARINI

1.Introduction271

2.ChemometricStrategiesAppliedinDataFusion273

3.Building,Optimization,andValidationofData-FusedModels276

4.Applications278

5.Conclusions301 References301

11.ImageFusion

ANNADEJUAN,AOIFEGOWEN,LUDOVICDUPONCHEL,ANDCYRILRUCKEBUSCH

1.Introduction311

2.ImageFusionbyUsingSingleFusedDataStructures314

3.ImageFusionbyConnectingDifferentImagesThroughRegressionModels323

CONTENTS vii

Detection,Classification,andImageLibrarySearching JOHNH.KALIVAS

References341
Acknowledgments368 References368 Index371
viii
4.ImageFusionforSuperresolutionPurposes328 5.Conclusions340
12.DataFusionofNonoptimizedModels:ApplicationstoOutlier
1.OutlierDetection346 2.Classification356 3.ThermalImageAnalysis364
CONTENTS

Contributors

D.Ballabio MilanoChemometricsandQSARResearchGroup,Departmentof EarthandEnvironmentalSciences,UniversityofMilanoBicocca,Milano,Italy

AlessandraBiancolillo DepartmentofChemistry,UniversityofRome“La Sapienza”Rome,Italy

LionelBlanchet

DepartmentofPharmacologyandToxicology,NUTRIMSchool forNutrition,andTranslationalResearchinMetabolism,MaastrichtUniversity, Maastricht,TheNetherlands

RicardBoque ´ UniversitatRoviraiVirgili,DepartmentofAnalyticalChemistry andOrganicChemistry,CampusSesceladesTarragona,Spain

LutgardeBuydens RadboudUniversity,InstituteforMoleculesandMaterials, DepartmentofAnalyticalChemistry,Nijmegen,TheNetherlands

V.Cariou StatSC,ONIRIS,INRA,Nantes,France

MarinaCocchi DepartmentofChemicalandGeologicalSciences,Universityof ModenaandReggioEmilia,Modena,Italy

V.Consonni MilanoChemometricsandQSARResearchGroup,Departmentof EarthandEnvironmentalSciences,UniversityofMilanoBicocca,Milano,Italy

AnnadeJuan ChemometricsGroup,Dept.ofChemicalEngineeringand AnalyticalChemistry,UniversitatdeBarcelona,Barcelona,Spain

L.DeLathauwer KULeuven,DepartmentofElectricalEngineeringESAT/ STADIUS,KasteelparkArenberg,Leuven,Belgium;GroupScience, EngineeringandTechnology,KULeuven-Kulak,Kortrijk,Belgium

LudovicDuponchel Universite ´ deLilleLASIR,Lille,France

JasperEngel Biometris,WageningenUniversityandResearch,Wageningen, TheNetherlands

AoifeGowen

SchoolofBiosystemsandFoodEngineering,UniversityCollege Dublin,Dublin,Ireland

D.Jouan-RimbaudBouveresse UMRInge ´ nierieProce ´ de ´ sAliments, AgroParisTech,Inra,Universite ´ Paris-Saclay,Massy,France;UMRPNCA, AgroParisTech,INRA,Universite ´ ParisSaclay,Paris,France

JohnH.Kalivas

DepartmentofChemistry,IdahoStateUniversity,Pocatello,ID, UnitedStates

FedericaMandreoli Dip.diScienzeFisiche,InformaticheeMatematiche, Modena,Italy

FedericoMarini

DepartmentofChemistry,UniversityofRome“LaSapienza” Rome,Italy

ix

ManuelaMontangero Dip.diScienzeFisiche,InformaticheeMatematiche, Modena,Italy

TormodNæs NofimaAS,Aas,Norway;QualityandTechnology,Department ofFoodScience,FacultyofLifeSciences,UniversityofCopenhagen, Frederiksberg,Denmark

E.M.Qannari StatSC,ONIRIS,INRA,Nantes,France

CyrilRuckebusch Universite ´ deLilleLASIR,Lille,France

D.N.Rutledge UMRInge ´ nierieProce ´ de ´ sAliments,AgroParisTech,Inra, Universite ´ Paris-Saclay,Massy,France

AgeK.Smilde BiosystemsDataAnalysis,SwammerdamInstituteforLife Sciences,UniversityofAmsterdam,Amsterdam,TheNetherlands

AgnieszkaSmolinska DepartmentofPharmacologyandToxicology,NUTRIM SchoolforNutrition,andTranslationalResearchinMetabolism,Maastricht University,Maastricht,TheNetherlands

EwaSzymanska FrieslandCampina,Amersfoort,TheNetherlands

R.Tauler IDAEA-CSIC,Barcelona,Spain

R.Todeschini MilanoChemometricsandQSARResearchGroup,Departmentof EarthandEnvironmentalSciences,UniversityofMilanoBicocca,Milano,Italy

IvenVanMechelen ResearchGrouponQuantitativePsychologyandIndividual Differences,KULeuven,Leuven,Belgium

N.Vervliet KULeuven,DepartmentofElectricalEngineeringESAT/STADIUS, KasteelparkArenberg,Leuven,Belgium

CONTRIBUTORS x

Preface

Thisbookdealswithdatafusionaimingatfurnishingavisionofthe differentavailablemethodologiesandthedataanalyticschallenges,framingatthesametimethenatureofcoupleddataandhowdatafusioncan enhanceknowledgediscovery.Tothisaim,thisbookwillalsofocuson methodsthatallowunravellingcommonanddistinctinformationfrom differentblocksofdata.

Theadoptionofdata-drivendiscoveryparadigminsciencehasledto theneedofhandlinglargeamountofdiversedata.Driversofthischange areononehandtheincreasedavailabilityandaccessibilityofhyphenated analyticalplatform,imagingtechniques,theexplosionofomicsdata,and ontheotherhandthedevelopmentofinformationtechnology.Hence,the mainchallengeishowtofacethesemultiplesourcedataandhowto retrieveallpossibleavailableinformation.Oneofthesalientaspectsis themethodologytointegratedatafrommultiplesources,analyticalplatforms,differentmodalities,varyingtimescale,aswellasunstructured data.Thisisgenerallyreferredtoasdatafusion.

Datafusionisforsureahotissue,aswellasamultifacetedconcept, and,asitemergesfromliterature,inrecentyearstherehasbeenaprogressivedevelopmentofawealthofmethods.However,themainattitudehas been,withfewexception,topresent/discussspecifictoolsconfinedina givendisciplineorcontext,especiallywithinthefieldofmonographs thefocuswasmostlyposedonremotesensingandmultisensordata integration.Aunifiedviewisstilllacking,whichpreventsauseful contaminationacrossdisciplines,awiderunderstandingandproperdisseminationofthemethodology.

Thisbookhastheambitiontoaddresstheseissues,providingacomprehensiveandcomprehensibledescriptionofthecurrentstateofthe artandofferingacross-disciplinaryapproachandaknowledgeretrieval basedperspectivetogetherwithpresentingchallengingandconvincing applications.

Thisbookismultiauthoredandwrittenasacollectionofindependent chaptersthatprovidedifferentperspectivesandapplications.However, whenconsideredaltogether,thechaptersintegrate,givingsoundbasis tounderstandthedatafusionprocess.

Thefirstfivechapterscoverthebasicconceptsandthemainmethodologies.Introductiondetailswaysandmeanstoaccomplishdatafusion andtherelatedtaxonomiesframingtheminauser’sperspective.

xi

AltogetherChapters2,3,and5provideageneralframeworkforlow-, mid-,andhigh-leveldatafusionmethodologiesincludingmotivating applicationsinlifescienceandanalyticalchemistry;inparticular,Chapter 2presentsagenericmodelforcoupleddatablocksaimedatrecovering fullinformationineachsingledatablock,aswellastheircommon,distinctiveinformationandthelinkingrelations.Chapter4presentsavery generalandflexiblemathematicalframework,whichallowsmatrices and/ortensorstobecoupledthrough(partially)sharedfactorsor throughcommonunderlyingvariables,addressingaswelltechniques suitabletohandlebigdata.

Chapters6,7,and8illustrate,inatutorialmanner,multiblockandmultisetmethodsintheperspectiveofdatafusion,showingthespecificities andlinkamongtheseapproaches.

Thelastfourchapters(Chapters9 12)aremoreorientedtowardapplications,suchasfoodcharacterizationandauthenticity(Chapter10), imageanalysis(Chapter11),and/orspecificambitsasChapter12,which presentan“ensemble”approachtodatafusionwithapplicationstooutlierdetection,classification,andimagelibrarysearching,andChapter9 whichillustratesthechallenges,andpossiblesolutionfromcomputerscience,indealingwithdataheterogeneity,e.g.,infusionofsemanticdatain lifesciencedomain.

Aconspicuousefforthasalsobeendevotedtopresentanextensivebibliographywhichisnonethelessincomplete,whenconsideringtheamount ofscientificliteratureonthesubject,butcouldserveasagoodlisttoface thesubject.

Writtenbyinvitedauthorswhoarerecognizedexpertsintheirfield, thisbookisaddressedtograduatestudents,researchers,andpractitioners;chaptersarewritteninawaytobeunderstandabletolargeand diverseaudienceandcontainenoughinformationtobe,byitself,sufficientandtobereadindependentlyoftheotherchapters.

Whatthereaderwouldfindisputtingdatafusionincontextandperspective,amultidisciplinaryview,detailedmethodsdescription,and challengingapplications.

Finally,Iwouldliketoexpressmysinceregratitudetoalltheinvited authorswhohaveacceptedtocooperateandcontributedtothisbook, aswellasmyapologiesforthedelayanddifficultiessomehowencountered.Ialsowishtothanktheseveralpeoplewho,onewayoranother, supportedthisproject.

PREFACE xii

Introduction:WaysandMeansto DealWithDataFromMultiple Sources

DepartmentofChemicalandGeologicalSciences,UniversityofModenaand ReggioEmilia,Modena,Italy

1.MOTIVATION

Theinterestindataintegrationorfusionhasgainedrenewedattention inrecentyearsowingtoboththecontinuousdevelopmentofinstrumental techniquesandsensordevicesandaparadigmchangeinthestudyof complexsystemstowardholistic,data-drivenapproaches [1,2].Ontheone hand,thetechnologicaldevelopmentininstrumentationhasincreasedthe dataacquisitionspeed,thecouplingofdifferentinstrumentalmodalities (hyphenation),theirportability(insitu,on/in-line),andthedatastorage capacity,thusleadingtoanenormousgrowthofavailabledatafor analysis;ontheotherhand,theparadigmshiftreliesondata-intensive statisticalexplorationanddataminingforknowledgediscovery.

Theneedfordataintegrationisthusbecomingubiquitousand encompassesseveraldifferentdisciplines.Roughlysummarizing,three mainareas,whichalsoidentifythemainreferencecommunities(listed distinctlyforbrevity,but,however,nottobeintendedasexclusive),can beevidenced:multisensors/remotesensing [3,4] (engineering,geoscience,signalprocessing),internetofthings/bigdata [5,6] (informatics, machinelearning,computerscience),andjointanalysisofmultipledata setsacquiredbydifferentmodalities(differenttypesofinstruments, experiments,andsettings) [7] (chemometrics,psychometrics,applied mathematics).

DataFusionMethodologyandApplications

https://doi.org/10.1016/B978-0-444-63984-4.00001-6

CHAPTER 1
1 Copyright © 2019ElsevierB.V.Allrightsreserved.

Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.