A New Approach for Android Malware Detection Based on Behavioral Profiling Techniques

The sharp increase in the number and types of Android malware poses a significant challenge to the development of effective Android malware detection methods. In this paper, we propose a new intelligent computing method to enhance the effectiveness of Android malware detection models. Our approach combines malware behavior profiling with the extraction of important features from these profiles using graph neural networks. Specifically, there are 3 main phases in the proposed model: (Phase 1) Behavioral Profiling: This phase proposes a new information enrichment technique based on new graph features and natural language processing method; (Phase 2) Feature Extraction: To extract features of Android malware, the paper proposes using an advanced graph neural network, GraphSAGE; (Phase 3) Classification: Based on the extracted features in Phase 2, the paper applies a Multi-layer Perceptron algorithm to classify malware and benign files. The approach to detecting Android malware in this paper can be applied to other problems, such as detecting URL malware, APT malware, and Botnet. The experimental results demonstrate that our proposed model outperforms other models across all metrics.