简介：Thispaperdescribessomeinertialalgorithmsfortheapproximationofnonlincarevolutionequationsandstudiesaprioriestimatesoftheapproximatesolutiomandthestabilityoftheinertialalgorithms.

简介：Inthispaper,wediscusstheconvergenceoftheBroydenalgorithmswithrevisedsearchdirection.Undersomeinexactllnesearches,weprovethatthealgorithmsaregloballyconvergentforcontinuouslydifferentiablefunctionsandtherateofconvergenceofthealgorithmsisone-stepsuperlinearandn-stepsecond-orderforuniformlyconvexobjectivefunctions.

简介：Superpixelsegmentationistheoversegmentationofanimageintoasetofhomogeneousregions.Superpixelhasmanyspecificpropertiesandhasbeencommonlyusedassupportingregionsforprimitivestoreducecomputationsinvariouscomputervisiontasks.Onepropertyofsuperpixelsiscompactness,whichispreferredinsomeapplications.Inthispaper,wegiveanreviewonimagesuperpixelsegmentationalgorithmsproposedinrecentyears.Superpixelsegmentationapproachesareclassifiedbasedonthecompactnessconstraintandtheirmainideaareintroduced.Wealsocomparethesealgorithmsinvisualandevaluatethemwithfivecommonmeasurements.

简介：

简介：

简介：WeconsideraclassofABStypealgorithmsforsolvingsystemoflinearinequalities,wherethenumberofinequalitiesdoesnotexceedthenumberofvariables.

简介：Iterativealgorithmsforsolvingthedataassimilationproblemsareconsidered，basedonthemainandadjointequations.Spectralpropertiesofthecontroloperatorsoftheproblemarestudied,theiterativealgorithmsarejustified.

简介：［1］BorgeforsG.Distancetransformationsinarbitrarydimensions.Computer,Vision,GraphicsandImageProcessing,1984(27):321～345［2］LarkinJ.AnANSICprogramtodetermineinexpectedlineartimetheverticesoftheConvexHullofasetofplanarpoints.Computers&Geosciences,1991,17(3):431～443［3］SidesE,HackR.SeminarReport:three-dimensionalGIS:futuredevelopments.ITCJournal,1995(2):151～164［4］FritschD.Three-dimensionalgeographicalinformationsystemsstatusandprospects.In:InternationalArchivesofPhotogrammetryandRemoteSensing,Vienna,1996,(B3):215～278［5］GongJY,XiaZG.Anintegrateddatamodelinthree-di-mensionalGIS.In:ProceedingsofGeoinformatics'96WuhanInternationalSymposium.1996.107～120［6］ManacedonioG,PareschiMT.Analgorithmforthetriangulationofarbitrarilydistributedpoints:applicationstovolumeestimateandterrainfitting.Computer&Geosciences,1991,17(7):859～874［7］MolenaarM.Atopologyfor3Dvectormaps.ITCJournal,1992(1):25～33［8］PiloukM,etal.ATetrahedron-based3DvectordatamodelforgeoinformationIn:AdvancedGeographicDataModeling.NetherlandsGeodeticCommission,PublicationsonGeodesy,1994.129～140［9］LiQQ,LiDR.Datastructuresinthreedimensionalgeographicinformationsystem.JournalofWuhanTechnicalUniversityofSurveyingandMapping,1996,21(2):128～133(inChinese)［10］LiQQ,LiDR.HybriddatastructurebasedonOctreeandTetrahedronin3-DGIS.In:InternationalArchivesofPhotogrammetryandRemoteSensing,Vienna,1996,31(B4):503～507［11］LiQQ,LiDR.Studyondatastructurein3-DGIS.In:ProceedingsofGeoinformatics'96WuhanInternationalSymposium,1997.229～237

简介：ThisletterintroducescolorconstancyandRetinextheoryforimageenhancement.Itclas-sifiesRetinexalgorithmsintofourcategoriesandprovidestheirprinciplesandimplementationsingeneral.TheexperimentalresultsofFrankle-McCann,MSR(Multi-ScaleRetinex)andPNSD(Pro-jectedNormalizedSteepestDescent)Retinexalgorithmsarepresentedandcompared.Moreover,varianceandaveragegradientareproposedtoevaluatetheperformanceofthedifferentalgorithms.

简介：Apiecewisecubiccurvefittingalgorithmpreservingmonotonicityofthedatawithoutmodificationoftheassignedslopesigproposed.ThealgorithmhasthesameorderofconvergenceasYan’salgorithm[8]andGasparo-Morandi’salgorithm[5]foraccurateorO(hq)accurategivendata,butithasamorevisuallypleasingcurvethanthosetwoalgorithms.WealsodiscusstheconvergenceorderofcubicrationalinterpolationforO(hq)accuratedata.

简介：ＣＵＲＶＥＡＮＤＳＵＲＦＡＣＥＩＮＴＥＲＰＯＬＡＴＩＯＮＢＹＳＵＢＤＩＶＩＳＩＯＮＡＬＧＯＲＩＴＨＭＳＲｕｉｂｉｎＱｕＣＵＲＶＥＡＮＤＳＵＲＦＡＣＥＩＮＴＥＲＰＯＬＡＴＩＯＮＢＹＳＵＢＤＩＶＩＳＩＯＮＡＬＧＯＲＩＴＨＭＳ￥ＲｕｉｂｉｎＱｕＡｂｓｔｒａ...

简介：

简介：在这份报纸，我们建议二新不安单一的变体。没有介绍人工的变量，解决线性编程问题，二的各个使用双枢规则完成最初的可行性，然后最初的枢规则完成optimality。第二个算法，修正第一，被设计更高效地处理高度退化的问题。有关不安的优点的一些有趣的结果被建立。从初步的测试的数字结果也被报导。[从作者抽象]

简介：Thispaperpresentsnonmonotonicquasi-Newtonalgorithmsviatwopre-conditionalcurvilinearpaths,thepreconditionalmodifiedgradientpathandtheprecon-ditionaloptimalpath,forunconstrainedoptimizationproblem.WeemploythestableBunch-Parlettfactorizationmethodtoformtwocurvilinearpathsveryeasily.Thenonmonotonecriterionisusedtospeeduptheconvergenceprogressinthecontoursofobjectivefunctionwithlargecurvature.Theoreticalanalysesaregivenwhichprovethattheproposedalgorithmsaregloballyconvergentandhavealocalsuperlinearcon-vergencerateundersomereasonableconditions.Theresultsofnumericalexperimentsarereportedtoshowtheeffectivenessoftheproposedalgorithms.

简介：尽管这被显示出HHT的实现--如果运动的方程被使用而不是差别方程评估下一步加速，一个方法能在假动态测试导致改进错误繁殖性质，当过去常解决一个非线性的系统时，这篇论文证明这个方法可能导致不稳定性。，它的无条件的稳定性为非线性的系统仅仅为线性有弹性的系统被验证，因为集中的步度是不到1，不稳定性发生。集中罐头的步度经常在假动态测试是不到1，是值得注意的，自从仅仅当集中的步度不管它的价值是否更大接近1时，一个会聚的答案被完成或不到1。因此，到假动态测试的这个计划的申请应该被禁止，自从不稳定性的可能性可能不正确地破坏一个标本。因而，由使用差别方程决定下一步加速的HHT-α方法的实现在假动态测试为使用被推荐。

简介：Inthispaper,theauthoratfirstdevelopsamethodtostudyconvergenceofthecascadealgorithminaBanachspacewithoutstableassumptionontheinitial(seeTheorem2.1),andthenappliesthepreviousresultontheconvergencetocharacterizingcompactlysupportedrefinabledistributionsinfractionalSobolevspacesandHoldercontinuousspaces(seeTheorems3.1,3.3,and3.4).Finallytheauthorappliestheabovecharacterizationtochoosingappropriateinitialtoguaranteetheconvergenceofthecascadealgorithm(seeTheorem4.2).

简介：为第二使用订Dini上面的方向性的衍生物的LC非强迫的优化问题介绍二个算法。方法的简洁使用并且表现；重复功能的相关性质的讨论。

简介：

简介：Thecross-fertilizationbetweenartificialintelligenceandcomputationalfinancehasresultedinsomeofthemostactiveresearchareasinfinancialengineering.Onedirectionistheapplicationofmachinelearningtechniquestopricingfinancialproducts,whichiscertainlyoneofthemostcomplexissuesinfinance.Intheliterature,whentheinterestrate,themeanrateofreturnandthevolatilityoftheunderlyingassetfollowgeneralstochasticprocesses,theexactsolutionisusuallynotavailable.Inthispaper,weshallillustratehowgeneticalgorithms(GAs),asanumericalapproach,canbepotentiallyhelpfulindealingwithpricing.Inparticular,wetesttheperformanceofbasicgeneticalgorithmsbyusingittothedeterminationofpricesofAsianoptions,whoseexactsolutionsisknownfromBlack-Scholesoptionpricingtheory.Thesolutionsfoundbybasicgeneticalgorithmsarecomparedwiththeexactsolution,andtheperformanceofGAsisewluatedaccordingly.Basedontheseewluations,somelimitationsofGAsinoptionpricingareexaminedandpossibleextensionstofutureworksarealsoproposed.

简介：AClassofCollinearScalingAlgorithmsforUnconstrainedOptimization.AnappealingapproachtothesolutionofnonlinearoptimizationproblemsbasedonconicmodelsoftheobjectivefunctionhasbeenintroducedbyDavidon(1980).Itleadstoabroadclassofalgorithmswhichcanbeconsideredtogeneralizetheexistingquasi-Newtonmethods.OneparticularmemberofthisclasshasbeendeeplydiscussedbySorensen(1980),whohasprovedsomeinterestingtheoreticalproperties.Inthispaper,wegeneralizeSorensen’stechniquetoSpedicatothree-parameterfamilyofvariable-metricupdates.Furthermore,wepointoutthatthecollinearscalingthree-parameterfamilyisessentiallyequivalenttotheSpedicatothree-parameterfamily.Inaddition,numericalexprimentshavebeencarriedouttocomparesomecollinerscalingalgorithmswithastraightforwardimplementationoftheBFGSquasi-Newtonmethod.