Comparative analysis of malicious Android-based software detection with trending metaheuristic algorithms

Abstract

Today, Android malware threats and attacks are rapidly increasing due to their use and popularity. Therefore, the need for systems effectively detecting malware is also increasing day by day. This study proposes the use of various trending metaheuristic algorithms for optimal feature selection (FS) in the detection of Android malware. For this purpose, the ten most prominent recent metaheuristic algorithms (RMAs) forfeature selection such as Artificial Bee Colony Algorithm (ABC), Firefly Algorithm (FA), Grey Wolf Optimisation (GWO), Ant Lion Optimisation (ALO), Crow Search Algorithm (CSA), Sine Cosine Algorithm (SCA), Whale Optimisation Algorithm (WOA), Salp Swarm Algorithm (SSA), Harris Hawk Optimization (HHO) and Butterfly Optimization Algorithm (BOA) were used for feature selection in this study. The efficiency of these algorithms is evaluated with five different machine learning (ML) methods on two well-known datasets of Android applications (Drebin215 and Malgenome-215). The results obtained are also compared with five well-known and widely used conventional metaheuristic algorithms (CMAs) for solving this problem. Extensive experimental results show that incorporating RMA into Android malware detection is a valuable approach.