简介:Manyreal-worldsystemscanbemodeledbyweightedsmall-worldnetworkswithhighclusteringcoefficients.Recentstudiesforrigorouslyanalyzingtheweightedspectraldistribution(WSD)havefocusedonunweightednetworkswithlowclusteringcoefficients.Inthispaper,werigorouslyanalyzetheWSDinadeterministicweightedscale-freesmall-worldnetworkmodelandfindthattheWSDgrowssublinearlywithincreasingnetworkorder(i.e.,thenumberofnodes)andprovidesasensitivediscriminationforeachinputofthismodel.ThisstudydemonstratesthatthescalingfeatureoftheWSDexistsintheweightednetworkmodelwhichhashighandorder-independentclusteringcoefficientsandreasonablepower-lawexponents.
简介:ItisshowninthispaperthatahyponormalweightedshiftofnormoneisunitarilyequivalenttoaToeplitzoperatorifandonlyifitsweights{an}0∞satisfy1-丨an丨2=(1-丨a0丨2)n+1n≥0.Inparticular,thisanswerstheAbrahamse’sProblem2.Asaconse-quence,the(surprising)answer,obtainedfirstbyC.C.Coweninanexplicitform,toHalmos’Question5isrecaptured.
简介:Let(Ω,A,P)beaprobabilityspace,X(t,ω)arandomfunctioncontinuousinprobabilityfort∈[0,∞)or(-∞,+∞)(ω∈Ω),andF(t)apositivefunctioncontinuousfort∈[0,+∞)or(-∞,+∞).IfX(t,ω)andF(t)verifycertainconditions,thenthereexistaasequence{Qn(t,ω)}ofrandompolynomialssuchthatwehavealmostsurely:fort[0,+∞)or(-∞,+∞),lim↑n→+∞|X(t,ω)-Qn(t,ω)|/F(t)=0.
简介:Theaimofthispaperistodevelopanorderedweighteddistance(OWD)measure,whichisthegeneralizationofsomewidelyuseddistancemeasures,includingthenormalizedHammingdistance,thenormalizedEuclideandistance,thenormalizedgeometricdistance,themaxdistance,themediandistanceandthemindistance,etc.Moreover,theorderedweightedaveragingoperator,thegeneralizedorderedweightedaggregationoperator,theorderedweightedgeometricoperator,theaveragingoperator,thegeometricmeanoperator,theorderedweightedsquarerootoperator,thesquarerootoperator,themaxoperator,themedianoperatorandtheminoperatorarealsothespecialcasesoftheOWDmeasure.SomemethodsdependingontheinputargumentsaregiventodeterminetheweightsassociatedwiththeOWDmeasure.TheprominentcharacteristicoftheOWDmeasureisthatitcanrelieve(orintensify)theinfluenceofundulylargeorundulysmalldeviationsontheaggregationresultsbyassigningthemlow(orhigh)weights.ThisdesirablecharacteristicmakestheOWDmeasureverysuitabletobeusedinmanyactualfields,includinggroupdecisionmaking,medicaldiagnosis,datamining,andpatternrecognition,etc.Finally,basedontheOWDmeasure,wedevelopagroupdecisionmakingapproach,andillustrateitwithanumericalexample.
简介:WestudytheboundedandthecompactweightedcompositionoperatorsfromtheBlochspaceintotheweightedBanachspacesofholomorphicfunctionsonboundedhomogeneousdomains,withparticularattentiontotheunitpolydisk.Forboundedhomogeneousdomains,wecharacterizetheboundedweightedcompositionoperatorsanddeterminetheoperatornorm.Inaddition,weprovidesufficientconditionsforcompactness.Fortheunitpolydisk,wecompletelycharacterizethecompactweightedcompositionoperators,aswellasprovide'computable'estimatesontheoperatornorm.
简介:Inthispaper,weproposetwoweightedlearningmethodsfortheconstructionofsinglehiddenlayerfeedforwardneuralnetworks.Bothmethodsincorporateweightedleastsquares.Ourideaistoallowthetraininginstancesnearertothequerytoofferbiggercontributionstotheestimatedoutput.Byminimizingtheweightedmeansquareerrorfunction,optimalnetworkscanbeobtained.Theresultsofanumberofexperimentsdemonstratetheeffectivenessofourproposedmethods.
简介:ThereseearchontherelationbetweenweightedshiftoperatorsonHilbertspacesandotherimportantclassofoperatorsattractedtheattentionofsomemathematicians.Forexample,therelationbetweenweightedshiftoperatorsandsubnormaloperatorshasbeenthoroughlystudiedbyJ.Stampfli,R.GellarandD.A.Herrero,etc.(seereference[1])Butthedecomposabilityofweightedshiftoperatorshasnotyetattractedenoughattentionuptonow.Wemadeinitialresearch
简介:Theweightedreliabilityofnetworkisdefinedasthesumofthemultiplicationoftheprobabilityofeachnetworkstatebyitsnormalizedweightingfactor.Underacertainstate,whenthecapacityformsourcestosinktislargerthanthegivenrequiredcapacityCr,thenthenormalizedweightingfactoris1,otherwise,itistheratioofthecapacitytotherequiredcapacityCr.Thispaperproposesanewalgorithmfortheweightedreliabilityofnet-works,putsforwardtheconceptofsaturatedstateofcapacity,andsuggestsarecursiveformulaforexpandingtheminimalpathstobethesumofqualifyingsubsets.Inthenewalgorithm,theexpansionsoftheminimalpathsdon'tcreatetheirrelevantqualifyingsubsets,thusdecreasingtheunnecessaryexpandingcalculation.Comparedwiththecurrentalgorithms,thisalgorithmhastheadvantageofasmallamountofcomputationsforcomputerimplementa-tion.
简介:这份报纸被奉献在同类的组X上学习加权的强壮的空间的原子分解。一些结果是在欧几里德几何学的泛音分析学习的对象的扩展。
简介:在网络的社区察觉在最后十年广泛地被学习了。许多标准,表示分区的质量获得,以及一些准确算法和很多启发规则被建议了。吝啬标准在最小化增加或从给定的网络搬迁了以便把它转变成一套disjoint派系的边的数字在于。最近,张,丘和张建议了一个重量系数在被介绍平衡插入并且删除的边的数字的一个加权的吝啬模型。这些作者建议规则选择系数的好值,使用模仿了退火发现最佳或在最佳附近的解决方案并且解决一系列真实、人工的例子。在现在的纸,一个算法为确切为参数的所有价值解决加权的吝啬问题被建议。这个算法反复地基于用一个排产生算法为参数的一套给定的价值解决这个问题。这个过程与一个搜索过程被相结合发现值曲线的所有最低断点即,插入并且删除的边的加权的和。从文学的一系列人工、真实的世界网络上的计算结果被报导。看来,为一样的几个分区联网可能增进知识并且解决方案的集合通常包含至少一个直觉地呼吁的分区。
简介:<正>SupposeXisasuperdiffusioninRdwithgeneralbranchingmechanism¢.andYDdenotesthetotalweightedoccupationtimeofXinaboundedsmoothdomainD.WediscusstheconditionsonψtoguaranteethatYDhasabsoluteycontinuousstates.Andforparticularψ(z)=z(l+,0DisabsolutelycontinuouswithrespecttotheLebesguemeasureinD.whereasinthecased>2+2/B.itissingular.AsweknowtheabsolutecontinuityandsingularityofY(Dhavenotbeendiscussedbefore.
简介:MostexistingapplicationsofcentroidalVoronoitessellations(CVTs)lackconsiderationofthelengthoftheclusterboundaries.Inthispaperweproposeanewmodelandalgorithmstoproducesegmentationswhichwouldminimizethetotalenergy—asumoftheclassicCVTenergyandtheweightedlengthofclusterboundaries.TodistinguishitwiththeclassicCVTs,wecallitanEdge-WeightedCVT(EWCVT).TheconceptofEWCVTisexpectedtobuildamathematicalbaseforallCVTrelateddataclassificationswithrequirementofsmoothnessoftheclusterboundaries.TheEWCVTmethodiseasyinimplementation,fastincomputation,andnaturalforanynumberofclusters.
简介:在这篇论文,我们在联合起来的磁盘D上学习Hakopian插值的加权的吝啬的不可分的集中。我们出现在Hakopian插值多项式Hn之间的内部产品(f;x,y);光滑的功能g(x,y)在D上收敛到f的(x,y);g(x,y)在D上什么时候n→8f(x,y)属于C(D);g的所有第一部分衍生物(x,y)属于空间嘴唇M[α](0<α≤1)。我们进一步证明那提供了g的所有秒部分的衍生物(x,y)也属于空间嘴唇M[α];f(x,y)属于C[1](D),在Hakopian插值的部分衍生物之间的内部产品多项式;g(x,y)在D上收敛到那在之间;g(x,y)在D上什么时候n→∞。