Bio-inspired for Features Optimization and Malware Detection

摘要: The leaking of sensitive data on Android mobile device poses a serious threat to users, and the unscrupulous attack violates privacy users. Therefore, an effective malware detection system is necessary. However, detecting challenging due similarity permissions in with those seen benign applications. This paper aims evaluate effectiveness machine learning approach for malware. In this paper, we applied bio-inspired algorithm as feature optimization selecting reliable permission features that able identify attacks. A static analysis technique classifier developed from noted shows use potential detection. study compares [particle swarm (PSO)] evolutionary computation information gain find best features. were optimized 378 11 by using algorithm: particle (PSO). evaluation utilizes 5000 Drebin samples 3500 samples. recognizing malware, it appears AdaBoost achieve good accuracy true positive rate value 95.6%, permissions. results show (PSO)