简介：High-frequencystocktrendpredictionusingmachinelearnershasraisedsubstantialinterestinliterature.Nevertheless,thereisnogoldstandardtoselecttheinputsforthelearners.Thispaperinvestigatestheapproachofadaptiveinputselection(AIS)forthetrendpredictionofhigh-frequencystockindexpriceandcomparesitwiththecommonlyuseddeterministicinputsetting(DIS)approach.TheDISapproachisimplementedthroughcomputationoftechnicalindicatorvaluesondeterministicperiodparameters.TheAISapproachselectsthemostsuitableindicatorsandtheirparametersforthetime-varyingdatasetusingfeatureselectionmethods.Twostate-of-the-artmachinelearners,supportvectormachine(SVM)andartificialneuralnetwork(ANN),areadoptedaslearningmodels.AccuracyandF-measureofSVMandANNmodelswithboththeapproachesarecomputedbasedonthehigh-frequencydataofCSI300index.TheresultssuggestthattheAISapproachusingt-statistics,informationgainandROCmethodscanachievebetterpredictionperformancethantheDISapproach.Also,theinvestmentperformanceevaluationshowsthattheAISapproachwiththesamethreefeatureselectionmethodsprovidessignificantlyhigherreturnsthantheDISapproach.

简介：Theproblemof“strungballs”,whichconsidershowmanycirclesinagivenfamilyofcirclescanbeintersectedbyalineinaplane,isdiscussedinR^n(n≥3).Higherdimensionalversionsoftheresultsgivenonaplaneareobtained.

简介：Inthispaper,thelinearmixedfiniteelementmethodisconsideredforthebiharmonicDirichletboundaryvalueproblem.Almostoptimalerrorestimates,superconvergenceestimatesandthederivativeerrorexpansionareobtained.

简介：

简介：Inmostexitingportfolioselectionmodels,securityreturnsareassumedtohaverandomorfuzzydistributions.However,uncertaintiesexistinactualfinancialmarkets.Marketsareassociatednotonlywithinherentrisk,butalsowithbackgroundriskthatresultsfromthedifferencesamongindividualinvestors.Thispaperinvestigatedthecomplianceofstockyieldstothefuzzy-naturedhigh-ordermomentsofrandomnumbersinordertodevelopahigh-momenttrapezoidalfuzzyrandomportfolioriskmodelbasedonvariance,skewness,andkurtosis.DataobtainedfromtheShanghaiStockExchangeandShenzhenStockExchangewasusedtoassesstheinfluenceontheproposedmodelofbothbackgroundriskandthemaximumlevelofsatisfactionoftheportfolio.Theempiricalresultsdemonstratedthatthedifferencesbetweenthemaximumandminimumvariance,skewness,andkurtosisvaluesoftheportfoliowerepositivelycorrelatedwiththevarianceofthebackgroundrisk.

简介：

简介：Lowgainfeedbackreferstocertainfamiliesofstabilizingstatefeedbackgainsthatareparameterizedinascalarandgotozeroasthescalardecreasestozero.Lowgainfeedbackwasinitiallyproposedtoachievesemi-globalstabilizationoflinearsystemssubjecttoinputsaturation.Itwasthencombinedwithhighgainfeedbackindifferentwaysforsolvingvariouscontrolproblems.Theresultingfeedbacklawsarereferredtoaslow-and-highgainfeedback.Sincetheintroductionoflowgainfeedbackinthecontextofsemi-globalstabilizationoflinearsystemssubjecttoinputsaturation,therehasbeenefforttodevelopalternativemethodsforlowgaindesign,tocharacterizekeyfeaturesoflowgainfeedback,andtoexplorenewapplicationsofthelowgainandlow-and-highgainfeedback.Thispaperreviewsthedevelopmentsinlowgainandlow-and-highgainfeedbackdesigns.

简介：ByapplyingtwononlinearGrangercausalitytestingmethodsandrollingwindowstrat-egytoexploretherelationshipbetweenspeculativeactivitiesandcrudeoilprices,theunidirectionalGrangercausalityfromspeculativeactivitiestoreturnsofcrudeoilpricesduringthehighpricephaseisdiscovered.ItisprovedthatspeculativeactivitiesdidcontributetohighcrudeoilpricesaftertheAsianfinancialcrisisandOPEC'soutputcutin1998.TheunidirectionalGrangercausalityfromreturnsofcrudeoilpricestospeculativeactivitiesissignificantingeneral.Butafter2000,withthesharpriseincrudeoilprices,thisunidirectionalGrangercausalitybecameacomplexnonlinearrelationship,whichcannotbedetectedbyanylinearGrangercausalitytest.

简介：Thispaperconsidersamulti-agenttrackingproblemforahigh-dimensionalactiveleaderandvariableinterconnectiontopology.Thestateoftheleadernotonlykeepschangingbutalsomaynotbemeasured.Toestimatethestatesuchaleaderindividually,aneighbor-basedlocalcontrollertogetherwithaneighbor-basedstate-estimationruleisgivenforeachautonomousagent.Then,theauthorsprovethat,withthehelpofaconstructedcommonLyapunovfunction(CLF),eachagentcantracktheactiveleaderwithunmeasurablestates.Finally,theauthorsexplicitlyconstructaCLFforanactiveleaderwithunknownperiodicinputforillustration.

简介：这份报纸为基于F统计数值在高维的部分线性的模型测试回归系数建议一个测试过程。在部分线性的模型，作者首先由一些nonparametric方法估计未知非线性的部件然后概括F统计数值在一些常规条件下面测试回归系数。在这个过程期间，非线性的部件的评价带许多挑战探索概括F-test的性质。作者在更一般的盒子中获得概括F-test的一些asymptotic性质，包括asymptotic规度和有p/n的这测试的力量(0，1)没有规度假设。asymptotic结果是一般的，由增加一些限制条件，我们能在高维的线性模型获得类似的结论。通过模拟研究，作者与理论结果比较表明建议测试的好有限样品的性能。我们的方法的实际用途被一个真实数据例子说明。

简介：

简介：Forageneralsecond-ordervariablecoefficientellipticboundaryvalueprobleminthreedimensions,theauthorsderivetheweakestimateofthefirsttypefortensor-productlinearpentahedralfiniteelements.Inaddition,theestimatefortheW1,1-seminormofthediscretederivativeGreen’sfunctionisgiven.Finally,theauthorsshowthatthederivativesofthefiniteelementsolutionuhandthecorrespondinginterpolantΠuaresupercloseinthepointwisesenseoftheL∞-norm.