Risk Assessment of Android Malwares Using Machine Learning Techniques

Deepthi Naidu Padrithi, Tennessee State University


ABSTRACT As android is the widely used operating system for smartphones and mobile devices, due to its popularity it has been targeted by malware developers. Security of android mainly depends on user decisions upon installing applications by granting the requested permissions. So to prevent the malicious application installation, the user should make appropriate decisions which helps the security of android based device. Usually, android user might not understand the requested permissions completely, they simply grant the access to the requested permissions to proceed with the application installation. This leads to major concern of android malware attacks. To overcome this, one way of securing device is to evaluate the risk of the malicious application and alert the user based on the risk priority. This will help the user to remove the malware from their device. There exists many risk analysis techniques in the market like risk computation using probabilistic methods which provide unreliable risk value. In this paper, we presented a model for risk assessment of android applications. This system is developed using python and scikit-learn, a machine learning library for Python. The risk score has been calculated on 0 - 100 scale based on probability estimates of classifiers for a given app based on applications permissions.

Subject Area

Computer engineering|Information technology

Recommended Citation

Deepthi Naidu Padrithi, "Risk Assessment of Android Malwares Using Machine Learning Techniques" (2017). ETD Collection for Tennessee State University. Paper AAI10639974.