支持向量回归机参数优化的方法研究

《支持向量回归机参数优化的方法研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、万方数据AbstractTheperformanceofsupportvectorregression(SVR)hasavitalrelationshipwithparameterselection，butevennow,thereisnotadefinitivemathematicaltheorytoguidehowtochoosetheoptimalparameters．Forthepurposeofsolvingthisproblem，theparameterselectionofSVRistransformedtoacombinatorialoptimizationproblem．Th

2、entheoptimizationproblemsissolvedbybothofintelligentalgorithms．Aimingatthedeficiencyofartificialbeecolonyalgorithm(ABC)，theweightingfunction，thepresent．optimalfoodsourceandthechaoticsearchalgorithmareintroducedtoimprovetheupdateformsofABCandthesearchmethodofscoutbees．ThentheSVRmodelisproposedbasedon

3、improvedABC．Numericalexperimentationsindicatethattheimprovedalgorithmisfeasibilityandsuperiority．Takingtheshort—termtrafficflowdataasanexample，thepredictiveresultsoftheimprovedABC．SVRiScomparedwithACO—SVR、PSO—SVRandABC—SVR，theresultsindicatethattheimprovedABC—SVRissuperiortothepredictiveeffectoftheo

4、therthree，itsruntimeistheshortestandshowsgoodgeneralizationabilityandlearningability．Aimingatthedisadvantageofartificialfishswarmalgorithm(AFSA)withthelateslowconvergencespeedandlowprecision，theABCisintroducedtoimprovethesearchefficiencyofAFSAandtheAFSA—ABChybridalgorithmisobtained．ThenitisusedforSV

5、Rparameterokmization．NumericalexperimentationsshowthatthealgorithmCansolvetheproblemofparameteroptimization．ThemodelisappliedtopredicttheGDPofShanghaiCity,theresultsindicatethatthemodelissuperiorthantheexistingpredictionmodel，itprovidesanewwayforthepredictionofGDEKeyWords：supportvectorregression；par

6、ameteroptimization；kernelparameter；artificialbeecolonyalgorithm；artificialfishswarlTlalgorithm万方数据目录摘要⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯IAbstract⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯II1绪论⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯．11．1研究背景⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯11．2研究现状及意义⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯21．3研究内容及结构安排⋯⋯⋯⋯⋯⋯⋯⋯⋯

7、⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯32相关基础理论⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯．52．1支持向量回归机⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯．．52．1．1损失函数⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯．．52．1．2核函数⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯．62．1．3支持向量机的回归⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯⋯82．2人工蜂群算法⋯⋯⋯⋯⋯⋯