欢迎来到天天文库
浏览记录
ID:34705516
大小:2.80 MB
页数:37页
时间:2019-03-09
《钓鱼网页深度学习智能检测方法的研究》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、华北电力大学硕士学位论文摘要“钓鱼(Phishing)”网站攻击是指通过模仿真实站点,实现假冒合法者身份骗取用户信任,继而窃取个人账户或隐私信息的违法行为。随着钓鱼网站的危害越来越严重,钓鱼网页检测作为一种反钓鱼措施与技术受到普遍的关注和重视。本文提出一种基于深度信念网络(DeepBeliefNetworks,DBN)的钓鱼网页智能检测方法,在分析网页内容特征的基础上,用DBN模型分类检测识别钓鱼网页。首先,由于网页内容特征的提取在很大程度上影响着钓鱼网页的检测结果,论文全面提取了网页的文档特征和图像特征,并丰富了网页特征的种类,在对页面的HTML源码和DOM文档模
2、型进行分析的基础上,提取更加精准的钓鱼网页特征向量。其次,针对提取后网页文本和图像特征向量维数多且格式不一致的问题,论文采用流行学习中的Isomap算法对前面生成的网页特征向量进行规格化降维处理,使之符合钓鱼网页分类器的输入要求,并且提高检测精度。再次,论文运用深度学习的DBN算法,构建了钓鱼网页智能检测分类器,给出了基于DBN的钓鱼网页检测方法的具体步骤和实验结果,并分析了算法的优缺点。实验证明,论文提出的基于DBN的钓鱼网页检测方法有较高的检测精度和较低的误判率。最后,论文设计和实现了一个钓鱼网页的深度学习智能检测系统,系统采用C/S模式设计,并展示了系统架构设
3、计以及部分功能界面。关键词:钓鱼网页检测,网页特征提取,流行学习算法,深度信念网络华北电力大学硕士学位论文AbstractPhishingattacks,whichstealusers’accountinformationbyfakewebsites,havebecomeaseriousproblemontheInternet.Phishingwebsitesareforgedwebpagesthat;arecreatedbymaliciouspeopletomimicwebpagesofrealwebsitesanditattemptsto,defraudpeop
4、leoftheirpersonalinformation.DetectingandidentifyingPhishing1websitesisreallyacomplexanddynamicprobleminvolvingmanyfactorsandcriteria.Inthispaper,weproposeaphishingdetectionmechanismbasedonDeepBeliefNetworks(DBN),whichthefeaturesofwebimageandDOMobjects/pmpertiesofwebaretakeninconsidera
5、tion,andtheDeepBeliefNetworksisusedtodetectandclassifyphishingweb]pages.Firstly,thefeaturesofwebimageareextractedforcomplementingthedisadvantageofphishingdetectiononlybasedondocumentobjectmodel(DOM).Thenaccordingtotheresultsofwebimagesegmentation,thefeaturesofwebimagethatincludeboundar
6、yshape,黟ayhistogram,colorhistogram,andspatialrelationshipbetweensubgraphs..AccordingtoDOMobjects,thefeaturesofwebsensitive:informationareexamined,,0vhichincludeURL,form,SSLcertificateandSOon.Secondly,asthedataformatsthatcomprisewebcharacteristicvectoraredifferentandexistredundant,thewe
7、bclassifierneedsmoretimetohandlecharacteristicdata.Inordertosol。vetheproblem,thispaperintroducedtheefficiencyIsomaptonormalizeandreducedimensionofcharacteristic,cectors,thismakethedatatofitforthephishingwebclassifierinputrequirements.TheselNturesofwebpageareusedtoassessphishingpageby
此文档下载收益归作者所有