资源描述：
《Credit rating modelling by kernel-based approaches with supervised and semi-supervised learning》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、NeuralComput&Applic(2011)20:761–773DOI10.1007/s00521-010-0495-0EANN2009Creditratingmodellingbykernel-basedapproacheswithsupervisedandsemi-supervisedlearningPetrHa´jek•Vladimı´rOlejReceived:27January2010/Accepted:16November2010/Publishedonline:5December2010ÓSpringe
2、r-VerlagLondonLimited2010AbstractThispaperpresentsthemodellingpossibilitiesrepaymentofprincipalandinterestofanobligationisofkernel-basedapproachestoacomplexreal-worldprob-expressedbymeansofanumericalcreditrating.Thelem,i.e.corporateandmunicipalcreditratingclassific
3、ation.assessmentisrealizedbyaratingagency.CreditratingsmayBasedonamodeldesignthatincludesdatapre-processing,beconductedfornations,companies,municipalities,finan-thelabellingofindividualparametervectorsusingexpertcialinstitutions,bondofferings,andindividuals.Thejthk
4、nowledge,thedesignofvarioussupportvectormachinesratingclassxj[X,X={AAA,AA,…,D},inwhichXisawithsupervisedlearningaswellaskernel-basedapproachesratingscale,isaresultofacreditratingprocess.Theratingwithsemi-supervisedlearning,thismodellingisundertakenclassxj[Xisassig
5、nedtothei-thassessedobjectoi[O,inordertoclassifyobjectsintoratingclasses.TheresultsO={o1,o2,…,oi,…,on}.Basedontheabove,thecreditshowthattheratingclassesassignedtobondissuerscanberatingmodellingisconsideredtobeaclassificationproblemclassifiedwithhighclassificationaccu
6、racyusingalimitedwiththegoalofclassifyingobjectsoi[Ointoratingclassessubsetofinputvariables.Thisholdstrueforkernel-basedxj[X.approacheswithbothsupervisedandsemi-supervisedCreditratingsarecostlytoobtainduetothelargelearning.amountoftimeandhumanresourcesinvestedbyra
7、tingagenciestoperformthecreditratingprocess.ThereisaKeywordsCreditratingKernelSupportvectorgreatdealofeffortmadetosimulatethecreditratingmachinesSupervisedlearningSemi-supervisedlearningprocessofratingagenciesthroughstatistical(e.g.[31–33])andsoft-computingmet
8、hods(e.g.[7,24]).Thedifficultyindesigningsuchmodelsliesinthesubjectivityofthecredit1Introductionratingprocess.Thissubjectivityisaccentuatedasthepar-ticul
