简介:TheLS-SVM(Leastsquaressupportvectormachine)methodispresentedtosetupamodeltoforecasttheoccurrenceofthunderstormsintheNanjingareabycombiningNCEPFNLOperationalGlobalAnalysisdataon1.0°×1.0°gridsandcloud-to-groundlightningdataobservedwithalightninglocationsysteminJiangsuprovinceduring2007-2008.Adatasetwith642samples,including195thunderstormsamplesand447non-thunderstormsamples,arerandomlydividedintotwogroups,one(having386samples)formodelingandtherestforindependentverification.ThepredictorsareatmosphericinstabilityparameterswhichcanbeobtainedfromtheNCEPdataandthepredictandistheoccurrenceofthunderstormsobservedbythelightninglocationsystem.Preliminaryapplicationstotheindependentsamplesfora6-hourforecastofthunderstormeventsshowthatthepredictioncorrectionrateofthismodelis78.26%,falsealarmrateis21.74%,andforecastingtechnicalscoreis0.61,allbetterthanthosefromeitherlinearregressionorartificialneuralnetwork.
简介:SupportVectorMachine(SVM)isapowerfulmethodologyforsolvingproblemsinnon-linearclassification,functionestimationanddensityestimation,whichhasalsoledtomanyotherrecentdevelopmentsinkernelbasedmethodsingeneral.Thispaperpresentsahighaccuracyandfault-tolerantSVMforthemobilegeo-locationproblem,whichisanimportantcomponentofpervasivecomputing.Simulationresultsshowitsbasiclocationperformance,andillustrateimpactsofthenumberoftrainingsamplesandtrainingareaontestlocationerror.
简介:水泥回转窑熟料制作过程中主传动电机电流不稳定、波动范围大,文章结合粗糙集、最小二乘支持向量机原理对水泥回转窑主传动电流进行预测。首先介绍粗糙集、最小二乘支持向量机的原理,通过搜集影响水泥回转窑主传动电流变化的数据建立信息决策表并对其进行预处理,使用粗糙集对样本数据进行约简,包括属性约简、属性值约简,利用LS-SVM理论对约简后的数据进行处理及预测,并将其他数据用于训练测试,验证测试结果。融合后的方法克服了LS—SVM对冗余信息和关键信息识别的局限性,补偿RS理论对输入数据信息缺乏抗干扰能力的缺点,通过实验研究证明该方法有较强的泛化能力,且预测准确率高。
简介:摘要液气比是石灰石-石膏湿法烟气脱硫系统设计中的一个关键参数,对其进行精确预测有着非常重要意义。本文利用最小二乘支持向量机(LS-SVM)建立了液气比预测模型。测试样本集的液气比预测结果表明,LS-SVM液气比预测模型具有很好的预测性能。
简介:Inthiswork,somechemometricsmethodsareappliedforthemodelingandpredictionoftheHildebrandsolubilityparameterofsomepolymers.Ageneticalgorithm(GA)methodisdesignedfortheselectionofvariablestoconstructtwomodelsusingthemultiplelinearregression(MLR)andleastsquare-supportvectormachine(LS-SVM)methodsinordertopredicttheHildebrandsolubilityparameter.TheMLRmethodisusedtobuildalinearrelationshipbetweenthemoleculardescriptorsandtheHildebrandsolubilityparameterforthesecompounds.ThentheLS-SVMmethodisutilizedtoconstructthenon-linearquantitativestructure-activityrelationship(QSAR)models.TheresultsobtainedusingtheLS-SVMmethodarethencomparedwiththoseobtainedfortheMLRmethod;itwasrevealedthattheLS-SVMmodelwasmuchbetterthantheMLRone.Theroot-mean-squareerrorsofthetrainingsetandthetestsetfortheLS-SVMmodelwere0.2912and0.2427,andthecorrelationcoefficientswere0.9662and0.9518,respectively.ThispaperprovidesanewandeffectivemethodforpredictingtheHildebrandsolubilityparameterforsomepolymers,andalsorevealsthattheLS-SVMmethodcanbeusedasapowerfulchemometricstoolforthequantitativestructure-propertyrelationship(QSPR)studies.
简介:Itiswellknownthatwhentherandomerrorsareiid.withfinitevariance,theweekandthestrongconsistencyofLSestimateofmultipleregressioncoefficientsareequivalent.Thisnote,byconstructingacounter-example,showsthatthisequivalencenolongerholdstrueincasethattherandomerrorspossessonlyther-thmomentwith1≤r<2.
简介:Ahandgesturerecognitionmethodispresentedforhuman-computerinteraction,whichisbasedonfingertiplocalization.First,handgestureissegmentedfromthebackgroundbasedonskincolorcharacteristics.Second,featurevectorsareselectedwithequalintervalsontheboundaryofthegesture,andthengestures'lengthnormalizationisaccomplished.Third,thefingertippositionsaredeterminedbythefeaturevectors'parameters,andanglesoffeaturevectorsarenormalized.Finallythegesturesareclassifiedbysupportvectormachine.Theexperimentalresultsdemonstratethattheproposedmethodcanrecognize9gestureswithanaccuracyof94.1%.
简介:Microarray数据基于肿瘤诊断是在生物信息学的一个很有趣的话题。关键问题之一是一个肿瘤的增进知识的基因的发现和分析。尽管解决这个问题有许多精致的途径,仅仅与microarray数据为肿瘤诊断选择增进知识的基因的一个合理集合仍然是困难的。在这份报纸,我们分类经由敏感对手惩罚了竞争学习的距离(DSRPCL)通过microarray数据表示进很多簇的基因算法然后在支持向量机器(SVM)的帮助下检测增进知识的基因簇或集合。而且,批评或强大的增进知识的基因能在获得的增进知识的基因簇上通过进一步的分类和察觉被发现。它是我们的建议DSRPCL-SVM途径为肿瘤诊断导致增进知识的基因的一种合理选择的冒号,白血病,和乳癌数据集的实验表明的井。