Research Article
An Android Malware Detection Method Based on Optimized Feature Extraction Using Graph Convolutional Network
@INPROCEEDINGS{10.1007/978-3-031-56583-0_19, author={Zhiqiang Wang and Zhuoyue Wang and Ying Zhang}, title={An Android Malware Detection Method Based on Optimized Feature Extraction Using Graph Convolutional Network}, proceedings={Digital Forensics and Cyber Crime. 14th EAI International Conference, ICDF2C 2023, New York City, NY, USA, November 30, 2023, Proceedings, Part II}, proceedings_a={ICDF2C PART 2}, year={2024}, month={4}, keywords={Android Malware Graph Convolutional Networks Static Analysis Graph Features}, doi={10.1007/978-3-031-56583-0_19} }- Zhiqiang Wang
Zhuoyue Wang
Ying Zhang
Year: 2024
An Android Malware Detection Method Based on Optimized Feature Extraction Using Graph Convolutional Network
ICDF2C PART 2
Springer
DOI: 10.1007/978-3-031-56583-0_19
Abstract
With the development of the mobile Internet, mobile devices have been extensively promoted and popularized. Android, as the current popular mobile intelligent operating system, has encountered problems such as the explosive growth of Android malware while bringing convenience to users. The traditional Android malware detection methods have some problems, such as low detection accuracy and difficulty in detecting unknown malware. This paper proposes an Android malware detection method named Android malware detection method based on graph convolutional neural network (AGCN) based on the graph convolutional network (GCN) to solve the above problems. Firstly, we divide the Android software datasets according to family and software features and construct a directed network topology graph. At the same time, the permission features of APK files are extracted and vectorized. Then, we use GCN to learn the features of Android APK files… Finally, we compare AGCN with a multilayer perceptron (MLP), long and short-term memory (LSTM) neural network, bi-directional long and short-term memory (bi-LSTM) neural network, and deep confidence neural network (DCNN) for experiments. Experimental results show that the model has an accuracy of 98.55% for malware detection, demonstrating the detection method’s effectiveness.
