资源描述：
《Scalable Network Traffic Classification Using》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。
1、2015IEEE8thInternationalConferenceonCloudComputingScalableNetworkTrafficClassificationUsingDistributedSupportVectorMachinesDoLeQuoc∗,ValerioD’Alessandro†,ByungchulPark§,LuigiRomano‡,ChristofFetzer∗∗DresdenUniversityofTechnology,Dresden,GermanyEmail:{do.lequoc,christof.fetzer}@tu-dresden.de†University
2、ofNaplesFedericoII,Naples,ItalyEmail:vale.dalessandro@studenti.unina.it‡ParthenopeUniversityofNaples,Naples,ItalyEmail:luigi.romano@uniparthenope.it§UniversityofToronto,Toronto,CanadaEmail:byungchul.park@utoronto.caAbstract—Internettraffichasincreaseddramaticallyinre-Overthelastfiveyears,thegrowthoft
3、heInternettrafficcentyearsduetothepopularizationoftheInternetandthevolumedemonstratesthenotionofBigData.ThisimpedesappearanceofwirelessInternetmobiledevicessuchassmart-theapplicationofSVMsindataminingbecausetheSVMsphonesandtablets.TheexplosivegrowthofInternettraffichasintroducedapracticalexamplethatd
4、emonstratestheconceptruntimecouldscaleapproximatelycubicallywiththenumberofBigData.Accurateidentificationandclassificationoflargeofobservationsinthelargetrainingdataset[15].Moreover,thenetworktrafficdataplaysanimportantroleinnetworkman-largedatasetsdonotfitintomemoryandevenintheharddiskagementincluding
5、capacityplanning,networkforensics,QoSofasinglemachine.ToovercometheproblemsofBigData,andintrusiondetection.However,thestate-of-the-artsolutions,theparallelanddistributedmethodsforSVMtraininghavewhichrelyonadedicatedserver,arenotscalableforanalyzinghighvolumenetworktrafficdata.Inthispaper,weimplement
6、beenintensivelystudied.ToreducethetimespentinkerneladistributedSupportVectorMachines(SVMs)frameworkforSVMtrainingonlarge-scaledatasets,Grafetal.[9]introducedclassifyingnetworktrafficusingHadoop,anopen-sourcedis-CascadeSVM,amultilevelapproach.Intheapproach,thetributedcomputingframeworkforBigDataproce
7、ssing.Weoriginaltrainingdatasetispartitionedintosubsets.Then,andesignaglobalparameterstorethatmaintainstheglobalsharedSVMisusedtotraineachsubsetusingthesameparametersasparametersbetweenSVMtrainingnodes.Thed
