MaLwareClassificationUtilizingSupervisedLearninginAutonomousDrivingApplications

MaLwareClassificationUtilizingSupervisedLearninginAutonomousDrivingApplications.pdf
Modernvehiclesarevastlyemployingnewinformationtechnologies,whichprovidetremendousbenefitstovehiclesafetyandfueleconomy.However,theincreasingconnectivityalsomakesthevehiclevulnerabletopotentiallycyberattack.Theproblemofve-hiclemalwaredetectionandclassificationhasemergedasanissueforvehiclecybersecurity.AmalwaredatasetfromMicrosoftareana-lyzedrespecttotheclassfrequency,classescouplingeffectandfeatureimportance.Severalsupervisedlearningmethodsarecomparedbychangingthedatasetvolume.Afterthat,a3-levelhierarchicalmethodisproposedformalwareclassification.Thefirstlevelutilizesthir-teensinglemodelstoestimatethemalwareclasses,whichactastheinputtothesecondlevelmodels.Thesecondleveliscomposedofthreemodels,whichareselectedbasedontheperformanceofthefirstlevelmodels,whilethethirdlevelmodeltakesweightedpredic-tionfromthesecondlevelandgeneratesthefinalmalwareclassificationprediction.Theproposedmethodreducesthemalwareclassifica-tionloglossby25.7%comparingwiththebestsinglemodelandisabletoachieve99.4%classificationaccuracy.
作者：XuBin ZhangDarui TangShuxian XuJiaxiong
作者单位：BlackHoleBigDataInc.;ClemsonUniversity-ICARClemsonUniversity-ICARGACToyotaInc.BlackHoleBigDataInc.
母体文献：第19届亚太汽车工程年会暨2017中国汽车工程学会年会论文集
会议名称：第19届亚太汽车工程年会暨2017中国汽车工程学会年会  
会议时间：2017年10月24日
会议地点：上海
主办单位：中国汽车工程学会
语种：chi
分类号：
关键词：cybersecurity  supervisedlearning  classification  autonomousdriving  automotive
在线出版日期：2020年7月9日
基金项目：
相似文献
相关博文

MaLwareClassificationUtilizingSupervisedLearninginAutonomousDrivingApplications.pdf

文件大小:

1008.2 KB

高速下载