欢迎来到天天文库
浏览记录
ID:41462979
大小:1.79 MB
页数:27页
时间:2019-08-25
《Real Life Applications of Soft Computing 5》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、3ArtificialNeuralNetworksII3.1tyPeSoFartiFiCialneuralnetWorKSChapter2providedanin-depthanalysisoftheANNmodelanditstraininginthesupervisedmode.Thechapterintroducedthebackpropagationalgorithm(BPA)formachinelearninganddefinedvariousparameterscommonlyusedinANNterminology.Theideaspresentedin
2、thatchapterdenotejustoneofthemanytypesofANNsthatareactivelyusedinvariousproblems.Theabundantresearchinthefieldofneurocomputing,aswellastheactiveinputsfromassoci-atedfieldssuchasthecognitivesciences,hasledtovariousANNmodelsandlearningtechniques.WhereasChapter2offeredanin-depthanalysisofp
3、roblemsolvingwithANNandBPA,thischap-terbroadlyexploresseveralothermodelsandmethodologiestointroduceyoutotheinfinitepos-sibilitiesandreallifeapplicabilityofANN.Themajormodelsstudiedinthischapterareradialbasisfunctionnetworks,learningvectorquantization,recurrentneuralnetworks,Hopfieldneur
4、alnetworks,andKohonen’sself-organizingmaps.Eachofthesemodels,initsownway,mapsinputstooutputswiththehelpofinferencesfromhistoricaldatabasesbymachinelearning.Westudiedthesupervisedlearninginthepreviouschapter.Inthischapterwepresenttheothertwotypesoflearning,i.e.,theunsupervisedlearningand
5、reinforcementlearning.Theselearningarethebasisofeveryneuralnetworkmodelthatwestudy.3.1.1unSuperviSedlearningInunsupervisedlearning(UL),thesystemisgivenonlytheinputstolearn;theoutputisunknowntothesystem.Therefore,thesystemhastolearnortrainitselfwithoutknowingthefinaloutcome,whichiswhyaUL
6、systemdoesnotlearnaboutitsperformanceandtheerrors.Thistypeoflearningisverydifferentfromsupervisedlearning,inwhichthewholelearningwasbasedontheoutput—inotherwords,theoutputdrovethesystemalthoughtrue.Itmaybehardtoimaginethatthesystemisbeingtrainedwithouthavingthefinalanswers.Thus,unsuperv
7、isedlearningcanbedefinedasthetechniqueofmachinelearningwhentheinputsareknownandtheoutputsareunknown.Definition27Unsupervisedlearningisthetechniqueofmachinelearningwhentheinputsareknownandtheoutputsareunknown.ULhasadeepanalogytothenaturalsystems.Muchofthelearninginhumanbeingsisi
此文档下载收益归作者所有