摘要
针对当前移动终端使用中存在的安全隐患,研究了一种新的面向Android移动终端的入侵检测算法。首先是在Android平台上收集移动终端内核信息并进行预处理,通过引入快速核密度估计(fast kernel density estimation,Fast KDE)算法对收集到的大规模样本进行压缩,得到数量合理的训练样本,然后结合在线增量学习算法,利用支持向量机(SVM)算法对处理后的数据进行判别以识别出入侵。实验结果表明,该方法极大缩短了训练时间,检测性能逐步达到最佳,具有较好的可扩展性和自提升能力。
In order to solve hidden security risks of mobile terminal, this paper proposed a new intrusion detection algorithm for Android mobile terminal. Firstly, the proposed system normalized kernel information, which was collected on the Android platform. And it obtained a reasonable number of training samples by introducing fast kernel density estimation algorithm （FastKDE）. Based on incremental learning online algorithm, using support vector machine （SVM） which was good at han- dling classification of small sample data, and the system determined whether it was invaded or not. The experimental results show that this method greatly shortens the training time, and gradually achieves the best detection performance, with better scalability and self-enhancing capabilities.
出处
《计算机应用研究》
CSCD
北大核心
2015年第9期2774-2778,共5页
Application Research of Computers
基金
江苏省自然科学基金重点项目(BK2011003)
国家自然科学基金资助项目(61103223)
关键词
Android移动终端
入侵检测
快速核密度估计
支持向量机
在线学习
Android mobile terminal
intrusion detection
fast kernel density estimation
support vector machine （ SVM ）
online learning
