Neural-Network-1-(神經(jīng)網(wǎng)絡(luò)自學(xué)的英文材料)_第1頁
Neural-Network-1-(神經(jīng)網(wǎng)絡(luò)自學(xué)的英文材料)_第2頁
Neural-Network-1-(神經(jīng)網(wǎng)絡(luò)自學(xué)的英文材料)_第3頁
Neural-Network-1-(神經(jīng)網(wǎng)絡(luò)自學(xué)的英文材料)_第4頁
Neural-Network-1-(神經(jīng)網(wǎng)絡(luò)自學(xué)的英文材料)_第5頁
已閱讀5頁,還剩49頁未讀, 繼續(xù)免費(fèi)閱讀

下載本文檔

版權(quán)說明:本文檔由用戶提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請進(jìn)行舉報(bào)或認(rèn)領(lǐng)

文檔簡介

IntroductionCourseObjectives

Thiscoursegivesanintroductiontobasicneuralnetworkarchitecturesandlearningrules.Emphasisisplacedonthemathematicalanalysisofthesenetworks,onmethodsoftrainingthemandontheirapplicationtopracticalengineeringproblemsinsuchareasaspatternrecognition,signalprocessingandcontrolsystems.WhatWillNotBeCoveredReviewofallarchitecturesandlearningrulesImplementationVLSIOpticalParallelComputersBiologyPsychologyHistoricalSketchPre-1940:vonHemholtz,Mach,Pavlov,etc.Generaltheoriesoflearning,vision,conditioningNospecificmathematicalmodelsofneuronoperation1940s:Hebb,McCullochandPittsMechanismforlearninginbiologicalneuronsNeural-likenetworkscancomputeanyarithmeticfunction1950s:Rosenblatt,WidrowandHoffFirstpracticalnetworksandlearningrules1960s:MinskyandPapertDemonstratedlimitationsofexistingneuralnetworks,newlearningalgorithmsarenotforthcoming,someresearchsuspended1970s:Amari,Anderson,Fukushima,Grossberg,KohonenProgresscontinues,althoughataslowerpace1980s:Grossberg,Hopfield,Kohonen,Rumelhart,etc.ImportantnewdevelopmentscausearesurgenceinthefieldApplicationsAerospaceHighperformanceaircraftautopilots,flightpathsimulations,aircraftcontrolsystems,autopilotenhancements,aircraftcomponentsimulations,aircraftcomponentfaultdetectorsAutomotiveAutomobileautomaticguidancesystems,warrantyactivityanalyzersBankingCheckandotherdocumentreaders,creditapplicationevaluatorsDefenseWeaponsteering,targettracking,objectdiscrimination,facialrecognition,newkindsofsensors,sonar,radarandimagesignalprocessingincludingdatacompression,featureextractionandnoisesuppression,signal/imageidentificationElectronicsCodesequenceprediction,integratedcircuitchiplayout,processcontrol,chipfailureanalysis,machinevision,voicesynthesis,nonlinearmodelingApplicationsFinancialRealestateappraisal,loanadvisor,mortgagescreening,corporatebondrating,creditlineuseanalysis,portfoliotradingprogram,corporatefinancialanalysis,currencypricepredictionManufacturingManufacturingprocesscontrol,productdesignandanalysis,processandmachinediagnosis,real-timeparticleidentification,visualqualityinspectionsystems,beertesting,weldingqualityanalysis,paperqualityprediction,computerchipqualityanalysis,analysisofgrindingoperations,chemicalproductdesignanalysis,machinemaintenanceanalysis,projectbidding,planningandmanagement,dynamicmodelingofchemicalprocesssystemsMedicalBreastcancercellanalysis,EEGandECGanalysis,prosthesisdesign,optimizationoftransplanttimes,hospitalexpensereduction,hospitalqualityimprovement,emergencyroomtestadvisementApplicationsRoboticsTrajectorycontrol,forkliftrobot,manipulatorcontrollers,visionsystemsSpeechSpeechrecognition,speechcompression,vowelclassification,texttospeechsynthesisSecuritiesMarketanalysis,automaticbondrating,stocktradingadvisorysystemsTelecommunicationsImageanddatacompression,automatedinformationservices,real-timetranslationofspokenlanguage,customerpaymentprocessingsystemsTransportationTruckbrakediagnosissystems,vehiclescheduling,routingsystemsBiology?Neuronsrespondslowly –10-3scomparedto10-9sforelectricalcircuits?Thebrainusesmassivelyparallelcomputation

–?1011neuronsinthebrain

–?104connectionsperneuronNeuronModelandNetworkArchitecturesSingle-InputNeuronTransferFunctionsTransferFunctionsMultiple-InputNeuronAbreviatedNotationLayerofNeuronsAbbreviatedNotationWw11,w12,?w1R,w21,w22,?w2R,wS1,wS2,?wSR,=b12S=bbbpp1p2pR=aa1a2aS=MultilayerNetworkAbreviatedNotationHiddenLayersOutputLayerDelaysandIntegratorsRecurrentNetworkAnIllustrativeExampleApple/BananaSorterPrototypeVectorsPrototypeBananaPrototypeAppleShape:{1:round;-1:eliptical}Texture:{1:smooth;-1:rough}Weight:{1:>1lb.;-1:<1lb.}MeasurementVectorPerceptronTwo-InputCaseDecisionBoundaryApple/BananaExampleThedecisionboundaryshouldseparatetheprototypevectors.Theweightvectorshouldbeorthogonaltothedecisionboundary,andshouldpointinthedirectionofthevectorwhichshouldproduceanoutputof1.ThebiasdeterminesthepositionoftheboundaryTestingtheNetworkBanana:Apple:“Rough”Banana:HammingNetworkFeedforwardLayerForBanana/AppleRecognitionRecurrentLayerHammingOperationInput(RoughBanana)FirstLayerHammingOperationSecondLayerHopfieldNetworkApple/BananaProblemTest:“Rough”Banana(Banana)SummaryPerceptronFeedforwardNetworkLinearDecisionBoundaryOneNeuronforEachDecisionHammingNetworkCompetitiveNetworkFirstLayer–PatternMatching(InnerProduct)SecondLayer–Competition(Winner-Take-All)#Neurons=#PrototypePatternsHopfieldNetworkDynamicAssociativeMemoryNetworkNetworkOutputConvergestoaPrototypePattern#Neurons=#ElementsineachPrototypePatternPerceptronLearningRuleLearningRules?SupervisedLearning

Networkisprovidedwithasetofexamples ofpropernetworkbehavior(inputs/targets)?ReinforcementLearning

Networkisonlyprovidedwithagrade,orscore, whichindicatesnetworkperformance?UnsupervisedLearning

Onlynetworkinputsareavailabletothelearning algorithm.Networklearnstocategorize(cluster) theinputs.PerceptronArchitectureWw11,w12,?w1R,w21,w22,?w2R,wS1,wS2,?wSR,=wiwi1,wi2,wiR,=WwT1wT2wTS=Single-NeuronPerceptronDecisionBoundary? Allpointsonthedecisionboundaryhavethesameinnerproductwiththeweightvector.? Thereforetheyhavethesameprojectionontotheweightvector,andtheymustlieonalineorthogonaltotheweightvectorExample-ORORSolutionWeightvectorshouldbeorthogonaltothedecisionboundary.Pickapointonthedecisionboundarytofindthebias.Multiple-NeuronPerceptronEachneuronwillhaveitsowndecisionboundary.Asingleneuroncanclassifyinputvectorsintotwocategories.Amulti-neuronperceptroncanclassifyinputvectorsinto2Scategories.LearningRuleTestProblemStartingPointPresentp1tothenetwork:Randominitialweight:IncorrectClassification.TentativeLearning

溫馨提示

  • 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
  • 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
  • 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁內(nèi)容里面會(huì)有圖紙預(yù)覽,若沒有圖紙預(yù)覽就沒有圖紙。
  • 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
  • 5. 人人文庫網(wǎng)僅提供信息存儲(chǔ)空間,僅對用戶上傳內(nèi)容的表現(xiàn)方式做保護(hù)處理,對用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對任何下載內(nèi)容負(fù)責(zé)。
  • 6. 下載文件中如有侵權(quán)或不適當(dāng)內(nèi)容,請與我們聯(lián)系,我們立即糾正。
  • 7. 本站不保證下載資源的準(zhǔn)確性、安全性和完整性, 同時(shí)也不承擔(dān)用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。

最新文檔

評論

0/150

提交評論