机器学习：局部与整体的学习(英文版),金国庆等著,浙江大学出版社

图书标准信息

• 作者 黄开竹 著
• 出版社 浙江大学出版社
• 出版时间 2008-04
• 版次 1
• ISBN 9787308058315
• 定价 70.00元
• 装帧 精装
• 开本 16开
• 纸张 胶版纸
• 页数 169页
• 正文语种 英语

【内容简介】

Machinelearning-modelingdatalocallyandgloballypresentsanovelandunifiedtheorythattriestoseamlesslyintegratedifferentalgorithms。specifically，thebookdistinguishestheinnernatureofmachinelearningalgorithmsaseither“locallearning”or“globallearning。”thistheorynotonlyconnectspreviousmachinelearningmethods，orservesasroadmapinvariousmodels，butmoreimportantlyitalsomotivatesatheorythatcanlearnfromdatabothlocallyandglobally。thiswouldhelptheresearchersgainadeeperinsightandcomprehensiveunderstandingofthetechniquesinthisfield。thebookreviewscurrenttopics，newtheoriesandapplications。
kaizhuhuangwasaresearcheratthefujitsuresearchanddevelopmentcenterandiscurrentlyaresearchfellowinthechineseuniversityofhongkong。haiqinyangleadstheimageprocessinggroupathisilicontechnologies。irwinkingandmichaelr。lyuareprofessorsatthecomputerscienceandengineeringdepartmentofthechineseuniversityofhongkong。

【目录】

1introduction
1.1learningandglobalmodeling
1.2learningandlocalmodeling
1.3hybridlearning
1.4majorcontributions
1.5scope
1.6book0rganization
references
2globallearningvs.locallearning
2.1problemdefinition
2.2globallearning
2.2.1generativelearning
2.2.2non—parametriclearning
2.2.3theminimumerrorminimaxprobabilitymachine
2.3locallearning
2.4hybridlearning
2.5maxi—minmarginmachine
references
3ageneralgloballearningmodehmempm
3.1marshalland0lkintheory

.3.2minimumerrorminimaxprobabilitydecisionhyperplane
3.2.1problemdefinition
3.2.2interpretation
3.2.3specialcaseforbiasedclassifications
3.2.4solvingthemempmoptimizationproblem
3.2.5whentheworst—casebayesoptimalhyperplanebecomesthetrueone
3.2.6geometricalinterdretation
3.3robustversion
3.4kernelization
3.4.1kernelizationtheoryforbmpm
3.4.2notationsinkernelizationtheoremofbmpm
3.4.3kernelizationresults
3.5experiments
3.5.1modelillustrationonasyntheticdataset
3.5.2evaluationsonbenchmarkdatasets
3.5.3evaluationsofbmpmonheart.diseasedataset
3.6howtightisthebound
3.7ontheconcavityofmempm
3.8limitationsandfuturework
3.9summary
referencese
4learninglocallyandglobally：maxi-minmarginmachine
4.1maxi—minmarginmachine
4.1.1separablecase
4.1.2connectionswithothermodels
4.1.3nonseparablecase
4.1.4furtherconnectionwithminimumerrorminimaxprobabilitymachine
4.2boundontheerrorrate
4.3reduction
4.4kernelization
4.4.1foundationofkernelizationform4
4.4.2kernelizationresult
4.5experiments
4.5.1evaluationsonthreesynthetictoydatasets
4.5.2evaluationsonbenchmarkdatasets
4.6discussionsandfuturework
4.7summary
references
5extensionⅰ：bmpmforimbalancedlearning
5.1introductiontoimbalancedlearning
5.2biasedminimaxprobabilitymachine
5.3learningfromimbalanceddatabyusingbmpm
5.3.1fourcriteriatoevaluatelearningfromimbalanceddata
5.3.2bmpmformaximizingthesumoftheaccuracies
5.3.3bmpmforrocanalysis
6extensionⅱ：aregressionmodelfromm4
7extensionⅲ：variationalmarginsettingswithinlocaldata
8conclusionandfuturework
references
index