An efficient combined deep neural network based malware detection framework in 5G environment

While Android smartphones are widely used in 5G networks, third-party application platforms facing a rapid increase the screening of applications for market launch. However, on one hand, due to receipt excessive listing, review requires lot time and computing resources. On other multi-selectivity features, it is difficult determine best feature combination as criterion distinguishing benign malicious software. To address these challenges, this paper proposes an efficient malware detection framework based deep neural network called DLAMD that can face large-scale samples. An designed, which combines pre-detection phase detection. The package (APK) analyzed detail, permissions opcodes distinguish from quickly extracted APK. Besides, obtain subset attributes most, random forest with good effect selected importance selection convolutional (CNN) automatically hidden pattern inside features selection. In experiment, real data shared collection download verify high efficiency proposed method. results show comprehensive classification index F1-score reach 95.69%.