An Android Malware Detection Technique Based on Optimized Permissions and API

As the number of smartphone users increases in terms of billions every year, and users now store personal and sensitive information on their mobile device which gives a huge platform for a hacker to steal users sensitive information. We have proposed a method to detect android malware using permissions and API. We have generated two types of feature vector named as common and combined feature vector. We obtained 97.25% accuracy for common and 96.56% accuracy for combined features using logistic regression. Further, to reduce training and testing time of classification we have optimized the feature to 131 by removing low variance features with which we have achieved 95.87% accuracy.