An Android Malware Detection Approach Using Community Structures of Weighted Function Call Graphs

摘要: With the development of code obfuscation and application repackaging technologies, an increasing number structural information-based methods have been proposed for malware detection. Although, many offer improved detection accuracy via a similarity comparison specific graphs, they still face limitations in terms computation time need manual operation. In this paper, we present new method that automatically divides function call graph into community structures. The features these structures can then be used to detect malware. Our reduces by improving Girvan–Newman algorithm using machine learning classification instead subgraphs. To evaluate our method, 5040 samples 8750 benign were collected as experimental data set. evaluation results show is higher than three well-known anti-virus software two previous control flow graph-based families. runtime performance exhibits clear improvement over GN structure generation.