資源描述：
《problem statement》由會員上傳分享，免費(fèi)在線閱讀，更多相關(guān)內(nèi)容在教育資源-天天文庫。
1、DataMining40DataMiningApproachesforNetworkIntrusionDetectionFinalReport11/9/2006Group9KarlaBracamonteJeffreyGawlinskiJordanHarstadOmarRodriguezMichaelWrightDataMining40AbstractOverthepastyears,muchinteresthasbeenshowsindataminingtodetectnetworkintrusions.Thispaperprovides
2、methodologyandthoughprocesswithinthediscussedtask.Baseduponexperiencesinsuchafield,dataminingtechniquesarefurthersuggestedandwillvaryuponexpertiseandnetworkinfrastructure.Thispaperisintendedfortheusebycomputerandnetworksecurityprofessionalswhowishtostudyandlearnmoreaboutt
3、hescienceofdatamining,aswellasresearchwaysexpertsusecurrentdataminingmaterialaswaystofurthertheirknowledgeinintrusiondetection.DataMining40ProblemStatementThegoalofIntrusionDetectionSystems(IDS)istodetectanintrusionasithappensandbeabletorespondtoit.Aprimaryconcernwhenimpl
4、ementingIDS,especiallyintheformofDataMiningistheresultof“falsepositives.”Afalsepositivecouldbeasituationwheresomethingabnormaloccurs,butisnotnecessarilyanintrusion.AhighpercentageoffalsepositivesmayrenderanIDSuselessandcauseuserrevolts.Anotherproblemisthatof“falsenegative
5、s”,inwhichanintrusionisactuallyoccurringbutitgoesundetectedbyIDSorIDSmisclassifiestheevent.ChallengesandMotivationConsequently,challengesmaypotentiallyarisewithinallintrusiondetectionmethods,especiallythatofdatamining.Thebroaderfieldofthesechallengesisreferredtoas“threata
6、nalysis”andthedefinitionofwhatitentails.Thus,theunderlyingneedthatmotivatessuchresearchistoprovidereliabilitytothosecompaniesandorganizationsthatdependondata.Bestdefined,threatanalysisisthestudyofknownandunknownpatterns,whichanalyzecertaindatatrafficpathsandtheirfollowing
7、characteristics.Statisticalandinformationstateresearchonanythingfromneighborhoodsandtheirdemographicscanbealsoincludedwithinthreatanalysis,thusmakingthecharacterizationprocessforintrusiondetectionquitebroadanddifficult.Oneofthebestmethodsinprovidingreliabilityofalldatatha
8、tistrackedistoreducefalsepositivesanderrors,whileincreasingconsumerconfidenceindatamanagementpro
