您好,欢迎访问三七文档
当前位置:首页 > 金融/证券 > 金融资料 > 商业银行监管的SOM神经网络的分类方法
200266:100026788(2002)0620026207SOM,(,300072):L,L,,,.:;;;:N945.11;F830.33:AaOff2siteCommercialBankingRegulationBasedSelf2organizingFeatureMapNeuralNetworksXIONGXiong,ZHANGWei(InstituteofSystemsEngineering,TianjinUniversity,Tianjin300072,China)Abstract:Beingthecoreoftheoff2sitecommercialbankingregulationsystems,theoff2sitebankingregulationsystemplayakeyroleintheentirebankingregulationsys2tem.Inthispaper,weregardtheoff2siteregulationasakindofpatternrecognition,andwepresentanewmodelbasedself2organizingfeaturemapneuralnetworks.Usingthismodel,weclassifiedcommercialbankfromthepracticedataandmadeacorre2spondinganalysisoftheresult.Theeffectivenessoftheproposedapproachisclearlydemonstratedbytheexperimentresults.Keywords:off2siteregulation;classification;neuralnetworks;self2organizingfea2turemap1,,,L,LL,,,,,LLL,LLa:2000209206:(79790130);(96-170):(1972-),,,;,,42,,,.©1995-2005TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.LLL21981(T.Kohonen)(Self2organizingFeatureMap,SOM)L,,,[1]L,,LL,,L,L,L,()(Bubble),,L:,,,,L,,,,L,,LL2.1,,,,L11,1L111,30L1L1XRk,W,YRNZY=WÝX,Ý726SOM©1995-2005TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.,EuclidZ:1.Wj(j=1,2,,p)j,,t=1Z2.Xk(k=1,2,,m):1)WjXiWg,Xk-Wg=minpõj=1Xk-WjZnEF:E-F=6ni=1(ei-fk)22)g,Ng(t)ZXi,$wij=A(t)ûxki-wijû,wij=wij+$wijA(t)t,ZA(t)=0.2[1-tö10000],t=1,2,3,,z(500z10000)xkikiZwijij,jNg(t)Z3)t,2)ZZ4),,Z,Zj,jZ.D,JD(i),V(D)Z,l,D=1,J1(1)=1,V(1)=1Xk(k=2,3,,m),q,V(k)=q,V(D),V(k)=V(Dn),Dn,i=i+1,JD(i)=k,,D=D+1,JD(1)=k,V(D)=qZ,,DZ2.2,:,4%-8%L1.0%-3.0%[2]L75%[4]L,,,2,L2.3,Z:,8220026$wø-ij0LgL3,5,2,3,4,5,6,7,8L©1995-2005TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.28%12%8%-12%4%-8%4%10080604010%20%10%-20%0-10%10080604090%90%80%-90%50%-80%50%10080604015%15%15%-25%25%-35%35%1008060402%2%2%-4%4%-6%6%10080604010%5%5%-10%10%-15%15%1008060405%5%5%-15%15%-35%35%1008060405%8%5%-8%2%-5%2%10080604025%45%25%-45%10%-25%10%100806040120%80%80%-120%120%-180%180%10080604030%0%0-30%30%-50%50%100806040926SOML©1995-2005TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.gL,g1,0,gL,bgLbg=1,Y-Wg=minpj=1Y-Wj0,Y-Wgminpj=1Y-Wj,L3,40,11,4011L,L34,6L3.1,:t,T,lin,lin=14,tF0.2T;lin=8,0.2TtF0.8T;lin=0,t0.8TZ1000,,,:1,27,14,28,16,15ZZ3.3151,3,712,4,6,8,10,12,16,18,20,22,28,30145,11,14,21,24,26277,13,15,25,32,33169,23,292819,27,31,34,,4Z40.071300.050270.066970.058400.243570.158930.822860.373800.893300.666930.07223608080808080808080100800.012840.040450.024770.125670.344460.003080.433630.336220.987631.012930.0389140806060606040808080600.046400.039880.029670.089400.274310.096900.786900.348050.876630.840510.0543660806080606060808080600.070510.057650.077570.072010.235610.120010.724380.394310.933330.749530.07301608080808080608080100800.054000.049330.059400.059660.160300.118430.730400.343430.900860.837100.08473608080808080608080801000.060250.044420.072120.074200.196270.089800.594550.226920.899421.010470.0600560808080806060608080803.2,Z0320026©1995-2005TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.Z,,40,60,80100Z,,,,,Z27Z2345,::,,Z:,Z:,Z:,,Z:,,Z:,Z3.35Z136SOM©1995-2005TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.67510.07590.05210.68350.06590.23680.16330.79520.33510.8110.70320.057320.01450.08630.52300.13800.2992-0.00040.31290.53411.2571.01070.073530.05280.10260.25130.05910.21750.14750.60900.52730.7030.71630.086440.07520.06920.83300.06700.18190.21100.81430.63460.9960.81030.07155-0.01240.03820.15910.12840.3429-0.04030.46770.29230.9780.41720.015960.07590.05210.68350.06590.23680.16330.79520.33510.8110.70320.05736Z611152131442751615100%4Z4,Z,,Z,Z,,Z,,Z,Z(73)2320026©1995-2005TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.Z,Z,,Z:[1]ShusakuTsumoto,HiroshiTanaka.ExtractionofdomainknowledgefromdatabasesbasedonroughsetTheory[A].ProceedingsofFifthIEEEInternationalConferenceonFuzzySystems[C],1996,2:748-754.[2]MrozekA.Roughsetsanddependencyanalysisamongattributesincomputerimplementationsofex2pertsinferencemodels[J].Int.Jour.Man2MachineStudies.1989,30(4):457-473.[3]DimitrasAI,SlowinskiR,SusmagaR,ZopounidisC.Businessfailurepredictionusingroughsets[J].EuropeanJournalofOperationalResearch.1999,114(2):263-280.[4]JacksonAJ,PawlakZ,LeClairSR.Roughsetsappliedtothediscoveryofmaterialsknowledge[J].JournalofAlloysandCompounds,1998,279(1):14-21.[5]HongTP,WangTT,WangSL,ChienBL.Learningacoveragesetofmaximallygeneralfuzzyrulesbyroughsets[J].ExpertSystemswithApplications.2000,19(2):97-103.[6]PawlakZ.RoughSets-theoreticalAspectsofReasoningaboutData[M].Boston.LondonDor2drecht:KluwerAcademicPublishers,1992.[7]AleksanderOhrn.DiscernibilityandRoughSetsinMedicine:ToolsandApplications[D].Norwe2gianUniversityofScienceandTechnology,Norway,1999:53.[8]IvoDuntsch,GuntherGediga,Uncertaintymeasuresofroughsetprediction[J].ArtificialIntelli2gence,1998,106(1):109-137.[9]IvoDuntsch,GuntherGediga.Roughianroughinformationanalysis(extendedabstract)[A],A.Sydow.Proceedings15thIMACSWorldCongress,Vol.4,WissenschaftandTechnik[C],Berlin,1997,631-636.[10]SlowinskiK,TaszeiaR,SlowinskiR.Sensitivityanalysisofroughclassification[J].Int.J.Man2MachineStu
本文标题:商业银行监管的SOM神经网络的分类方法
链接地址:https://www.777doc.com/doc-236642 .html