关键词: 安卓安全, 深度学习, 恶意软件, 机器学习

Abstract: In order to improve the detection efficiency and accuracy of Android malicious application, an Android malware detection system based on deep learning is proposed. Deep learning especially emphasizes on Android Security features, which combines static and dynamic features for classification. In order to improve the accuracy of malware detection, attribute subset selection analysis is used to reduce the dimensionality of fusion features. Then, the Stacked Denoising Autoencoders is used for classification. The SDADLDroid, a static and dynamic analysis system that detects the Android malware, is proposed. In order to recognize different Android malware, different kinds of clustering algorithms can be applied to compare the malware modeling capability. We use a dataset containing more than fifteen thousand real applications, including seven thousands malware applications. Experimental results show that the deep learning technique is especially suitable for Android malware detection than machine learning and can achieve a high level of 95.8% accuracy with real-world Android application sets.

Key words: Android security, deep learning, Android malware, machine learning