multiple organ retrieval and exchange MORE in Canada文档格式.docx

multiple organ retrieval and exchange MORE in Canada文档格式.docx

《multiple organ retrieval and exchange MORE in Canada文档格式.docx》由会员分享，可在线阅读，更多相关《multiple organ retrieval and exchange MORE in Canada文档格式.docx（29页珍藏版）》请在冰豆网上搜索。

Abstract

Proteomicsaimsatdeterminingthestructure,functionandexpressionofproteins.High-throughputmassspectrometry（MS）isemergingasaleadingtechniqueintheproteomicsrevolution.Thoughitcanbeusedtofinddisease-relatedproteinpatternsinmixturesofproteinsderivedfromeasilyobtainedsamples,keychallengesremainintheprocessingofproteomicMSdata.Multiscalemathematicaltoolssuchaswaveletsplayanimportantroleinsignalprocessingandstatisticaldataanalysis.Awavelet-basedalgorithmforproteomicdataprocessingisdeveloped.AMATLABimplementationofthesoftwarepackage,calledWaveSpect0,ispresentedincludingprocessingproceduresofstep-intervalunification,adaptivestationarydiscretewaveletdenoising,baselinecorrectionusingsplines,normalization,peakdetection,andanewlydesignedpeakalignmentmethodusingclusteringtechniques.ApplicationstorealMSdatasetsfordifferentcancerresearchprojectsinVanderbiltIngramCancerCentershowthatthealgorithmisefficientandsatisfactoryinMSdatamining.

ArticleOutline

1.Introduction

2.WaveletsandapplicationsinproteomicMSdataanalysis

2.1.WaveletsforMALDI-TOFMSData

2.2.Waveletdenoisingstrategy

2.3.Analysisonwaveletdomain

3.Method:

processingprocedures

3.1.Step-intervalunificationanddenoising

3.2.Baselinecorrectionandnormalization

3.3.Peakdetectionandalignment

4.Results

4.1.Peakselectionbasedmethodresults

4.2.Waveletcoefficientsanalysisresults

Acknowledgements

References

Purchase

$31.50

237

Networkdesigntechniquesusingadaptedgeneticalgorithms 

AdvancesinEngineeringSoftware,Volume32,Issue9,September2001,Pages731-744

MitsuoGen,RunweiCheng,ShumuelS.Oren

Inrecentyearswehaveevidencedanextensiveeffortinthedevelopmentofcomputercommunicationnetworks,whichhavedeeplyintegratedinhumanbeing'

severydaylife.Oneofimportantaspectsofthenetworkdesignprocessisthetopologicaldesignprobleminvolvedinestablishingacommunicationnetwork.However,withtheincreaseoftheproblemscale,theconventionaltechniquesarefacingthechallengetoeffectivelyandefficientlysolvethosecomplicatednetworkdesignproblems.Inthisarticle,wesummarizedrecentresearchworksonnetworkdesignproblemsbyusinggeneticalgorithms（GAs）,includingmultistageprocessplanning（MPP）problem,fixedchargetransportationproblem（fc-TP）,minimumspanningtreeproblem,centralizednetworkdesign,localareanetwork（LAN）designandshortestpathproblem.Alltheseproblemsareillustratedfromthepointofgeneticrepresentationencodingskillandthegeneticoperatorswithhybridstrategies.LargequantitiesofnumericalexperimentsshowtheeffectivenessandefficiencyofsuchkindofGA-basedapproach.

2.AdaptationofGAs

3.Multistageprocessplanningproblems

3.1.Representation

3.2.Geneticoperators

3.3.Evaluation

3.4.Example

4.Fixedchargetransportationproblem

4.1.Representation

4.2.Geneticoperators

4.3.Evaluationandselection

4.4.Examples

5.Minimumspanningtreeproblem

5.1.Representation

5.2.Feasibilitycondition

5.3.Geneticoperators

6.Centralizednetworkdesignproblem

7.Localareanetworkdesignproblem

8.Shortestpathproblem

8.1.Representation

8.2.Pathgrowthprocedure

8.3.Geneticoperators

8.4.Compromiseapproach

8.5.Evaluationandselection

8.6.OverallprocedureofGA

8.7.Examples

9.Conclusions

238

Areconfigurablecomputingframeworkformulti-scalecellularimageprocessing 

MicroprocessorsandMicrosystems,Volume31,Issue8,3December2007,Pages546-563

ReidPorter,JanFrigo,AlConti,NealHarvey,GarrettKenyon,MayaGokhale

Cellularcomputingarchitecturesrepresentanimportantclassofcomputationthatarecharacterizedbysimpleprocessingelements,localinterconnectandmassiveparallelism.ThesearchitecturesareagoodmatchformanyimageandvideoprocessingapplicationsandcanbesubstantiallyacceleratedwithReconfigurableComputers.Wepresentaflexiblesoftware/hardwareframeworkfordesign,implementationandautomaticsynthesisofcellularimageprocessingalgorithms.Thesystemprovidesanextremelyflexiblesetofparallel,pipelinedandtime-multiplexedcomponentswhichcanbetailoredthroughreconfigurablehardwareforparticularapplications.Themostnovelaspectsofourframeworkincludeahighlypipelinedarchitectureformulti-scalecellularimageprocessingaswellassupportforseveraldifferentpatternrecognitionapplications.Inthispaper,wewilldescribethesystemindetailandpresentourperformanceassessments.Thesystemachievedspeed-upofatleast100×

forcomputationallyexpensivesub-problemsand10×

forend-to-endapplicationscomparedtosoftwareimplementations.

2.Background

2.1.Typicalneighborhoodfunctions

2.2.Implementingneighborhoodfunctions

2.2.1.Dataparallel

2.2.2.Instructionparallel

3.Systemoverview

3.1.Networkspecificationfile

3.2.Networkparameterfile

4.Hardwareoverview

4.1.HardwareAPI

4.2.SoftwareAPI

5.Hardwarecomponents

5.1.Neighborhoodmemoryaccess

5.2.Neighborhoodfunctions

5.2.1.Convolution

5.2.2.Morphology

5.2.3.Threshold

5.3.Datasequencing

5.3.1.Downsampling

5.3.2.Streamsplitting

5.3.3.Streammixing

5.3.4.Upsampling

5.4.Parametermodule

6.Applicationcasestudies

6.1.ApplicationI

6.2.ApplicationII

7.Conclusion

$35.95

239

APHID:

Anarchitectureforprivate,high-performanceintegrateddatamining 

FutureGenerationComputerSystems,Volume26,Issue7,July2010,Pages891-904

JimmySecretan,MichaelGeorgiopoulos,AnnaKoufakou,KelCardona

Whiletheemergingfieldofprivacypreservingdatamining（PPDM）willenablemanynewdataminingapplications,itsuffersfromseveralpracticaldifficulties.PPDMalgorithmsarechallengingtodevelopandcomputationallyintensivetoexecute.DevelopersneedconvenientabstractionstosimplifytheengineeringofPPDMapplications.Theindividualpartiesinvolvedinthedataminingprocessneedawaytobringhigh-performance,parallelcomputerstobearonthecomputationallyintensivepartsofthePPDMtasks.ThispaperdiscussesAPHID（ArchitectureforPrivateandHigh-performanceIntegratedDatamining）,apracticalarchitectureandsoftwareframeworkfordevelopingandexecutinglargescalePPDMapplications.Atonetier,thesystemsupportssimplifieduseofclusterandgridresources,andatanothertier,thesystemabstractscommunicationforeasyPPDMalgorithmdevelopment.ThispaperoffersadetailedanalysisofthechallengesindevelopingPPDMalgorithmswithexistingframeworks,andmotivatesthedesignofanewinfrastructurebasedonthesechallenges.

2.Distributedandprivacypreservingdatamining

3.Background/relatedwork

3.1.Highperformanceparalleldataminingonclustersandgrids

3.2.Distributeddatamining

3.2.1.Agent-basedapproachesforDDM

3.3.DDMmiddleware

3.3.1.WebservicesforDDM

3.4.ArchitecturestosupportPPDM

4.PPDMpreliminaries

5.APHID

5.1.Developmentmodel

5.1.1.Programstructure

5.1.2.Sharedvariables

5.2.Mainexecutionlayer

5.3.High-performancecomputingservices

5.4.PPDMservices

5.5.Datamanagementbetweenlayersandparties

6.Examplealgorithmimplementation

6.1.Notation

6.2.Example:

privacypreservingNaï

veBayesclassifier

6.2.1.Securesum

6.2.2.Calculatingfrequencyofattributes

7.Performanceresults

8.Discussion

9.Conclusionsandfuturework

Vitae

$19.95

240

Perspectivesforprocesssystemsengineering—Personalviewsfromacademiaandindustry 

Computers&

ChemicalEngineering,Volume33,Issue3,20March2009,Pages536-550

Karsten-UlrichKlatt,WolfgangMarquardt

Processsystemsengineering（PSE）hasbeenanactiveresearchfieldforalmost50years.Itsmajorachievementsincludemethodologiesandtoolstosupportprocessmodeling,simulationandoptimization（MSO）.Mature,commerciallyavailabletechnologieshavebeenpenetratingallfieldsofchemicalengineeringinacademiaaswellasinindustrialpractice.MSOtechnologieshavebecomeacommodity,theyarenotadistinguishingfeatureofthePSEfieldanymore.Consequently,PSEhastoreassessandtorepositionitsfutureresearchagenda.Emphasisshouldbep