The widespread use of devices connected to Android systems in various areas of human life has made it an attractive target for bad actors. In this context, the development of mechanisms that can detect Android malware is among the most effective techniques to protect against various attacks. Feature selection is extremely to reduce the size of the dataset and improve computational efficiency while maintaining the accuracy of the performance model. Therefore, in this study, the five most widely used conventional metaheuristic algorithms for feature selection in the literature, such as Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Simulated Annealing (SA), Ant Colony Optimization (ACO) and Differential Evolution (DE), was used to select features that best represent benign and malicious applications on Android. The efficiency of these algorithms was evaluated on the Drebin-215 and MalGenome-215 dataset using five different machine learning (ML) method including Decision Tree (DT), K-Nearest Neighbour (KNN), Naive Bayes (NB), Random Forest (RF) and Support Vector Machine (SVM). According to the results obtained from the experiments, DE-based feature selection and RF classifier are found to have better accuracy. According to the findings obtained from the experiments, it was seen that DE-based feature selection and RF method had better accuracy rate.
Relevant Articles | Author | # |
---|
Article | Author | # |
---|