基于MSVM算法的Android恶意应用检测研究

RESEARCH ON ANDROID MALICIOUS APPLICATION DETECTION BASED ON MSVM ALGORITHM

搜集Android系统良性和恶意应用样本,提取17个特征值构造特征向量。应用样本自学习的方法学习得到模糊SVM分类器中多个核函数线性组合的权重。构造MSVM算法对Android恶意应用样本进行分类测试,并采用交叉验证。实验结果证明MSVM算法与一般的SVM相比取得较好的分类效果。

We collected the benign and malicious application samples of Android system.17 characteristic values were extracted to construct the eigenvector.The weight of multiple kernel functions linear combination in fuzzy SVM classifier was learned by using the method of sample self-learning.We constructed MSVM algorithm to classify the Android malicious application samples.The experimental results show that the MSVM algorithm has a better classification effect compared with the general SVM.