摘要
移动终端爆发式增长造成了恶意应用的大量出现,给用户的隐私安全和财产安全带来了巨大的危害.为提高Android应用恶意性检测的准确性,本文将卡方检验与基尼不纯度增量相结合获取更有价值的特征属性;并改进朴素贝叶斯算法提高Android应用恶意性判断的准确性.实验结果表明:新的特征处理方法能够有效提高检测性能;同时,改进后的朴素贝叶斯算法相比原始算法而言准确率有较大的提升.
The explosive growth of mobile terminals has produced endless malicious applications,bring on great harm to the security of users’ privacy and property.To solve this problem,a method based on chi-squared test and Gini impurity increment was proposed for more valuable features extraction and the Naive Bayes algorithm improvement,so as to improve the estimation accuracy of Android malevolence applications.Test shows that the new features processing method can improve the classification performance of algorithms.At the same time,the improved Naive Bayes algorithm can achieve higher accuracy than before.
作者
刘亚姝
王志海
李经纬
赵烜
文伟平
LIU Ya-shu;WANG Zhi-hai;LI Jing-wei;ZHAO Xuan;WEN Wei-ping(School of Computer and Information Technology,Beijing Jiaotong University,Beijing 100044,China;School of Electrical and Information Engineering,Beijing University of Civil Engineeringand Architecture,Beijing 100044,China;School of Electronics Engineering and ComputerScience,Peking University,Beijing 102600,China)
出处
《北京理工大学学报》
EI
CAS
CSCD
北大核心
2019年第3期290-294,共5页
Transactions of Beijing Institute of Technology
基金
国家重点研发计划资助项目(2018YFB0803604)
国家自然科学基金重点资助项目(U1736218)
国家自然科学基金面上资助项目(61672086)