Android malware detection via efficient application programming interface call sequences extraction and machine learning classifiers

Abstract Malware detection is an important task for the ecosystem of mobile applications (APPs), especially Android ecosystem, and vital to guarantee user experience APPs. There have been some exiting methods trying solve problem malware detection, but suffer from several defects, such as high time complexity mediocre accuracy, which seriously decrease practicability existing methods. To these problems, in this study, we propose a novel framework, where contribute efficient Application Programming Interface (API) call sequences extraction algorithm investigation different types classifiers. In API extraction, transforming function graph multigraph into directed simple graph, successfully avoids unnecessary repetitive path searching. We also pruning search, further reduces number paths be searched. Our greatly complexity. generate transition matrix classification features investigate three machine learning classifiers complete task. The experiments are performed on real‐world Packages (APKs), results demonstrate that our method significantly running produces accuracy.