MFENet: An 1D CNN-based Model for Automatic Feature Extraction and Fusion
Yang Li, Huijuan Zhu · 2020
As Android phones become the mainstream of smartphones, more and more malicious applications are being developed to attack the system and steal user information. Many researches based on machine learning can automatically detect Android malware currently. In spite of that, these methods require manual feature selection, which relies on experience and may loss major features. In this paper, we propose a tiny Android malware detection model named MFENet. This model is based on the one-dimensional convolutional neural network (1D CNN). The core of the model is MFEBlock, which can extract key features and fuse them automatically. This model got 94.62% accuracy on the test dataset, which achieves the outstanding results for Android malware detection on an open dataset.