Instance based security risk value estimation for Android applications

摘要: Abstract Android has emerged as the widest-used operating system for smartphones and mobile devices. Security of this platform mainly relies on applications (apps) installed by device owner since permissions sandboxing have reduced attack surface. antivirus programs detect known malware based their signature, but they cannot zero-day viruses. Therefore, estimating security risk could be helpful comparing selecting apps that are more likely to malicious or benign estimated values. systematic assistance making appropriate decisions can significantly improve Android-based Additionally, markets leverage risks recognize suspicious further analysis. In study, a new metric is introduced effective estimation untrusted apps. While previously proposed measurements features such function calls, our devised benefits from non-malicious app instances. The uses identified normal samples compute Thus, represented in feature space, each input app, using distances Moreover, increase metric's detection rate, an instance weighting schema suggested. Empirical evaluations various datasets show instance-based higher rates than score measurements.