資源描述：
《Fundamentals of Machine Learning》由會員上傳分享，免費在線閱讀，更多相關(guān)內(nèi)容在教育資源-天天文庫。
1、Chapter2FundamentalsofMachineLearning2.1LearningMethodsLearningisafundamentalcapabilityofneuralnetworks.Learningrulesarealgo-rithmsfor?ndingsuitableweightsWand/orothernetworkparameters.Learningofaneuralnetworkcanbeviewedasanonlinearoptimizationproblemf
2、or?ndingasetofnetworkparametersthatminimizethecostfunctionforgivenexamples.Thiskindofparameterestimationisalsocalledalearningortrainingalgorithm.Neuralnetworksareusuallytrainedbyepoch.Anepochisacompleterunwhenallthetrainingexamplesarepresentedtothenetw
3、orkandareprocessedusingthelearningalgorithmonlyonce.Afterlearning,aneuralnetworkrepresentsacom-plexrelationship,andpossessestheabilityforgeneralization.Tocontrolalearningprocess,acriterionisde?nedtodecidethetimeforterminatingtheprocess.Thecomplexityofa
4、nalgorithmisusuallydenotedasO(m),indicatingthattheorderofnumberof?oating-pointoperationsism.Learningmethodsareconventionallydividedintosupervised,unsupervised,andreinforcementlearning;theseschemesareillustratedinFig.2.1.xpandyparetheinputandoutputofthe
5、pthpatterninthetrainingset,?ypistheneuralnetworkoutputforthepthinput,andEisanerrorfunction.Fromastatisticalviewpoint,unsuper-visedlearninglearnsthepdfofthetrainingset,p(x),whilesupervisedlearninglearnsaboutthepdfofp(y
6、x).Supervisedlearningiswidelyusedi
7、nclassi?cation,approx-imation,control,modelingandidenti?cation,signalprocessing,andoptimization.Unsupervisedlearningschemesaremainlyusedforclustering,vectorquantization,featureextraction,signalcoding,anddataanalysis.Reinforcementlearningisusuallyusedin
8、controlandarti?cialintelligence.Inlogicandstatisticalinference,transductionisreasoningfromobserved,spe-ci?c(training)casestospeci?c(test)cases.Incontrast,inductionisreasoningfromobservedtrainingcasestogeneralrules,whicharethenappliedtothetestcases.Mach
9、inelearningfallsintotwobroadclasses:inductivelearningortransductivelearning.Inductivelearningpursuesthestandardgoalinmachinelearning,whichistoaccuratelyclassifytheentireinputspace.Incontrast,transductivelearningfocusesK.-L.DuandM.N.S.Sw
