Personalized Machine Learning_Julian McAuley文档文件分享.pdfVIP

PersonalizedMachineLearning

Everydayweinteractwithmachinelearningsystemsofferingindividualizedpredic-

tionsforourentertainment,socialconnections,purchases,orhealth.Theseinvolve

severalmodalitiesofdata,fromsequencesofclickstotext,images,andsocialinter-

actions.Thisbookintroducescommonprinciplesandmethodsthatunderpinthedesign

ofpersonalizedpredictivemodelsforavarietyofsettingsandmodalities.

Thebookbeginsbyrevising‘traditional’machinelearningmodels,focusingon

howtoadaptthemtosettingsinvolvinguserdata;thenpresentstechniquesbasedon

advancedprinciplessuchasmatrixfactorization,deeplearning,andgenerativemod-

eling;andconcludeswithadetailedstudyoftheconsequencesandrisksofdeploying

personalizedpredictivesystems.

Aseriesofcasestudiesindomainsrangingfrome-commercetohealthplushands-

onprojectsandcodeexampleswillgivereadersunderstandingandexperiencewith

large-scalereal-worlddatasetsandtheabilitytodesignmodelsandsystemsforawide

rangeofapplications.

julianmcauleyhasbeenaProfessorattheUniversityofCaliforniaSanDiego

since2014.PersonalizedMachineLearningisthemainresearchareaofhislab,

withapplicationsrangingfrompersonalizedrecommendationtodialog,healthcare,

andfashiondesign.Heregularlycollaborateswithindustryonthesetopics,including

Amazon,Facebook,Microsoft,Salesforce,andEtsy.Hisworkhasbeenselectedfor

severalawardsincludinganNSFCAREERaward,andfacultyawardsfromAmazon,

Salesforce,Facebook,andQualcomm,amongothers.

PersonalizedMachineLearning

JULIANMCAULEY

UniversityofCaliforniaSanDiego

UniversityPrintingHouse,CambridgeCB28BS,UnitedKingdom

OneLibertyPlaza,20thFloor,NewYork,NY10006,USA