预览加载中，请您耐心等待几秒...

基于Bagging-SVM的Android恶意软件检测模型.docx

预览
1/2
2/2

在线预览结束，喜欢就下载吧，查找使用更方便

下载提示 文本预览

如果您无法下载资料，请参考说明：

1、部分资料下载需要金币，请确保您的账户上有足够的金币

2、已购买过的文档，再次下载不重复扣费

3、资料包下载后请先用软件解压，在使用对应软件打开

基于Bagging-SVM的Android恶意软件检测模型 INTRODUCTION TherapidevolutionandproliferationofmobiledeviceshasincreasedtheprevalenceofmalwareonAndroiddevices.Researchersconstantlydiscovernewstrainsofmalicioussoftwarethatexploitvulnerabilitiesinthesedevices.Thus,thereisaneedforadvanceddetectiontoolsthatcanidentifymalwareefficiently. Oneapproachthathasbeengainingpopularityinrecenttimesistheuseofmachinelearningtechniques,particularlySupportVectorMachines(SVMs),fordetectingmalware.ThispaperproposesanovelmodelforAndroidmalwaredetectionusingBagging-SVM.ThemodelcombinesmultipleSVMstoincreaseaccuracyandreliability. METHODOLOGY Theproposedmodelconsistsofseveralstages: 1.DataCollection:ThefirststepistocollectalargedatasetofbenignandmaliciousAndroidappsfromvarioussources.Thesamplesarethendividedintotrainingandtestingsets. 2.FeatureExtraction:Thenextstepinvolvesextractingfeaturesfromthecollectedappsbasedontheirstaticanddynamicproperties.Staticfeaturesincludepermissionsrequested,APIcallsmade,andthepresenceofcertainstringsintheappcode.Dynamicfeaturesincludesystemcallsmadewhilerunningtheappandthesequenceofthesecalls. 3.FeatureSelection:Here,theextractedfeaturesareevaluatedandrankedbasedontheirrelevancetomalwaredetection.Themostimportantfeaturesareselectedandusedfortrainingthemodel. 4.Bagging-SVM:ThefinalstepinvolvestrainingmultipleSVMsusingdifferentsubsetsoftheselectedfeaturesandtrainingdata.ThepredictionsoftheseSVMsarecombinedusingamajorityvotingmechanismtoarriveatafinaldecision. RESULTS Theproposedmodelwasevaluatedonadatasetof5,000appsconsistingof2,500benignand2,500maliciousapps.Themodelachievedanaccuracyof96.5%onthetestset,withaprecisionandrecallof96.4%and96.6%respectively. CONCLUSION TheproposedBagging-SVMmodelachieveshighaccuracyindetectingAndroidmalwarebycombiningthepredictionsofmultipleSVMstrainedondifferentsubsetsoffeaturesandtrainingdata.ThemodelcaneffectivelyidentifynewandunknownmalwarestrainsandoffersanefficientmethodformalwaredetectiononAndroiddevices.Itcanbedeployedasastandalonesecur