您好,欢迎访问三七文档
当前位置:首页 > 机械/制造/汽车 > 综合/其它 > 机器学习简介ppt课件
MachineLearning:AnOverview石立臣1Outline•Whatismachinelearning(ML)•Typesofmachinelearning•Workflow•Popularmodels•Applications•Futures2WhatismachinelearningTrainingset(labelsknown)Testset(labelsunknown)f()=“apple”f()=“tomato”f()=“cow”3Whatismachinelearning•Definition–Machinelearningreferstoasystemcapableoftheautonomousacquisitionandintegrationofknowledge–MachinelearningisprogrammingcomputerstooptimizeaperformancecriterionusingexampledataorpastexperienceComputerDataAlgorithmProgramKnowledgeKnowledge(new)4Whatismachinelearning•Everymachinelearningalgorithmhasthreecomponents–Representation•Model(rules,statistics,instance;logic,KNN,SVM,DNN,…)–Evaluation•Performance(accuracy,mse,energy,entropy,…)–Optimization•Parameters–Combinatorialoptimization–Convexoptimization–Constrainedoptimization5Typesofmachinelearning•Supervisedlearning–Trainingdataincludesdesiredoutputs•Unsupervisedlearning–Trainingdatadoesnotincludedesiredoutputs•Semi-supervisedlearning–Trainingdataincludesafewdesiredoutputs•Reinforcementlearning–Rewardsfromsequenceofactions6Typesofmachinelearning•Supervisedlearning–Classification:discreteoutput–Regression:continuousoutputBias-variance7TrainingandValidationDataFullDataSetTrainingDataValidationDataIdea:traineachmodelonthe“trainingdata”andthentesteachmodel’saccuracyonthevalidationdata8Underfitting&OverfittingPredictiveErrorModelComplexityErroronTrainingDataErroronTestDataIdealRangeforModelComplexityOverfittingUnderfitting9Typesofmachinelearning•Unsupervisedlearning–Clustering–Dimensionalityreduction–Factoranalysis10Typesofmachinelearning•Semi-supervisedlearning–Clusteringorclassification11Typesofmachinelearning•Reinforcementlearning–Robot&control12WorkflowPredictionTrainingLabelsTrainingTrainingImageFeaturesImageFeaturesTestingTestImageLearnedmodelLearnedmodelSlidecredit:D.HoiemandL.Lazebnik13Workflow•Features14Workflow•Models–Logic,Rules–Statistical,Blackboxmodel•Static,dynamicmodel•Onlinelearning•Ensemblelearning15Workflow•ArchitectureModelFeatureHardware16Popularmodels•Linearmodel:logisticregression,lineardiscriminantanalysis,linearregression(withbasisfunction)17Popularmodels•Nearestneighbor–Feature&distance18Popularmodels•Supportvectormachine19Popularmodels•Artificialneuralnetwork20Popularmodels•Decisiontree21Popularmodels•Collaborativefiltering22Popularmodels•Hierarchicalclustering•K-means•Spectralclustering•Manifoldlearning23Popularmodels•Hiddenmarkovmodel•Conditionalrandomfields24Applications25Applications26Applications27Applications28Applications29Applications30Applications31Applications32Applications•Attention33Applications•Imageclassification34Applications35Applications•Brainmachineinterface36Applications37Applications38Applications39Applications40Applications•Indirectillumination–Regression41Applications•Indirectillumination–kd-tree42Applications•Thecoreisthedataset!!!•Others–features–model&optimization43Futures•Decision•Control•Knowledge•Prediction44
本文标题:机器学习简介ppt课件
链接地址:https://www.777doc.com/doc-7290337 .html