基于对比权限模式的恶意软件检测方法

A Malware Detection Method Based on Contrasting Permission Patterns

针对Android系统特有的权限管理机制,引入对比权限模式来表示恶意软件与正常软件的差异,进而提出一种组合分类器用于对Android恶意软件的自动检测.每一个对比权限模式被视为一个弱分类器,并为最终的结果投票.基于真实数据的实验表明,所提出的组合分类器在性能上优于传统的分类方法.

Exploiting the specific permission mechanism of Android system, we introduce contrasting permission patterns to represent the essential differences between malwares and clean applications and present an ensemble classifier for automatically detecting Android malwares. Each contrasting permission pattern is regarded as a weak classifier and votes for the final decision. Experiments on real applications demonstrate that the proposed classifier outperforms commonly used classifiers in Android malware detection.