简介:LetGbeap-seriesgroupandΩbeacompactsubgroupofG.Letλ(x,r)andλ,(x,r)beA-belp-poissontypekernelandproducttypekernelOnΩrespectively.Inthispaperwediscusstheap-proximationpropertiesofsuchkernels,givetheestimate5oftheirmoments,obtainthedirectandin-verseapproximationtheorems.
简介:TheNeighborhoodPreservingEmbedding(NPE)algorithmisrecentlyproposedasanewdimensionalityreductionmethod.However,itisconfinedtolineartransformsinthedataspace.Forthis,basedontheNPEalgorithm,anewnonlineardimensionalityreductionmethodisproposed,whichcanpreservethelocalstructuresofthedatainthefeaturespace.First,combinedwiththeMercerkernel,thesolutiontotheweightmatrixinthefeaturespaceisgottenandthenthecorrespondingeigenvalueproblemoftheKernelNPE(KNPE)methodisdeduced.Finally,theKNPEalgorithmisresolvedthroughatransformedoptimizationproblemandQRdecomposition.Theexperimentalresultsonthreereal-worlddatasetsshowthatthenewmethodisbetterthanNPE,KernelPCA(KPCA)andKernelLDA(KLDA)inperformance.
简介:Integralcollisionkerneliselucidatedusingexperimentalresultsfortitania,silicaandaluminananoparticlessynthesizedbyFCVDprocess,andtitaniasubmicronparticlessynthesizedinatubefurnacereactor.Theintegralcollisionkernelwasobtainedfromaparticlenumberbalanceequationbytheintegrationofcollisionratesfromthekinetictheoryofdilutegasesforthefree-moleculeregime,fromtheSmoluchowskitheoryforthecontinuumregime,andbyasemi-empiricalinterpolationforthetransitionregimebetweenthetwolimitingregimes.Comparisonshavebeenmadeonparticlesizeandtheintegralcollisionkernel,showingthatthepredictedintegralcollisionkernelagreedwellwiththeexperimentalresultsinKnudsennumberrangefromabout1.5to20.
简介:Receiveroperatingcharacteristic(ROC)curvesareoftenusedtostudythetwosampleprobleminmedicalstudies.However,mostdatainmedicalstudiesarecensored.UsuallyanaturalestimatorisbasedontheKaplan-Meierestimator.InthispaperweproposeasmoothedestimatorbasedonkerneltechniquesfortheROCcurvewithcensoreddata.Thelargesamplepropertiesofthesmoothedestimatorareestablished.Moreover,deficiencyisconsideredinordertocomparetheproposedsmoothedestimatoroftheROCcurvewiththeempiricalonebasedonKaplan-Meierestimator.ItisshownthatthesmoothedestimatoroutperformsthedirectempiricalestimatorbasedontheKaplan-Meierestimatorunderthecriterionofdeficiency.Asimulationstudyisalsoconductedandarealdataisanalyzed.
简介:TheFFDalgorithmisoneofthemostfamousalgorithmsfortheclassicalbinpackingproblem.Inthispaper,someversionsoftheFFDalgorithmareconsideredinseveralbinpackingproblems.Especially,twoofthemappliedtothebinpackingproblemwithkernelitemsareanalyzed.Tightworst-caseperformanceratiosareobtained.
简介:Hyperspectralimageprovidesabundantspectralinformationforremotediscriminationofsubtledifferencesingroundcovers.However,theincreasingspectraldimensions,aswellastheinformationredundancy,maketheanalysisandinterpretationofhyperspectralimagesachallenge.Featureextractionisaveryimportantstepforhyperspectralimageprocessing.Featureextractionmethodsaimatreducingthedimensionofdata,whilepreservingasmuchinformationaspossible.Particularly,nonlinearfeatureextractionmethods(e.g.kernelminimumnoisefraction(KMNF)transformation)havebeenreportedtobenefitmanyapplicationsofhyperspectralremotesensing,duetotheirgoodpreservationofhigh-orderstructuresoftheoriginaldata.However,conventionalKMNForitsextensionshavesomelimitationsonnoisefractionestimationduringthefeatureextraction,andthisleadstopoorperformancesforpost-applications.Thispaperproposesanovelnonlinearfeatureextractionmethodforhyperspectralimages.Insteadofestimatingnoisefractionbythenearestneighborhoodinformation(withinaslidingwindow),theproposedmethodexplorestheuseofimagesegmentation.Theapproachbenefitsbothnoisefractionestimationandinformationpreservation,andenablesasignificantimprovementforclassification.Experimentalresultsontworealhyperspectralimagesdemonstratetheefficiencyoftheproposedmethod.ComparedtoconventionalKMNF,theimprovementsofthemethodontwohyperspectralimageclassificationare8and11%.Thisnonlinearfeatureextractionmethodcanbealsoappliedtootherdisciplineswherehigh-dimensionaldataanalysisisrequired.
简介:§1.IntroductionandMainResultLet(X,F)beaJBrXR'-valuedvector.AssumethatwhenX=xisgiven,thereexistsaconditionaldensityofYtobedenotedbyf(y[x),whichisaBorel-measurablefunctionof(x,y).Notethatwedonotassumetheexistenceofadensityfunctionof(X,F).Let(X-i,fi),—,(Xn,Fn)bei.i.d.samplesof(X,F).Ourpurposeistoestimatef(y\x)basedonthesesamples.Thisisaninterestingprobleminviewofeitherpuretheoryorpracticalapplications.MotivatedbytheideasuggestedinkernelandNNestimationsinthetheoryofnonparametricregressionanddensityestimates,thefirstauthorproposesthefollowingtwoclassesofestimatorsoff(y\x):
简介:Amajordifficultyinmultivariablecontroldesignisthecross-couplingbetweeninputsandoutputswhichobscurestheeffectsofaspecificcontrollerontheoverallbehaviorofthesystem.Thispaperconsiderstheapplicationofkernelmethodindecouplingmultivariableoutputfeedbackcontrollers.Simulationresultsarepresentedtoshowthefeasibilityftheproposedtechnique.
简介:Thekernelbasedtrackinghastwodisadvantages:thetrackingwindowsizecannotbeadjustedefficiently,andthekernelbasedcolordistributionmaynothaveenoughabilitytodiscriminateobjectfromclutterbackground.Forboostingupthefeature'sdiscriminatingability,bothscaleinvariantfeaturesandkernelbasedcolordistributionfeaturesareusedasdescriptorsoftrackedobject.Theproposedalgorithmcankeeptrackingobjectofvaryingscalesevenwhenthesurroundingbackgroundissimilartotheobject'sappearance.
简介:Epicentral distribution in 1994Pei-ShanCHEN(陈培善)(InstituteofGeophysics,StateSeismologicalBureau,Beijing100081,China)Pei-ShanC...