Android Malicious Code Detection Strategy Based on Deep Learning
Yongliang Fu, Haihang Wang · 2025
With the widespread adoption of smartphones and the extensive use of the Android operating system, malware attacks have become a significant security issue, threatening users' privacy, data security, and device stability. Traditional malware detection methods largely rely on simple machine learning algorithms. However, as malware attack techniques become increasingly sophisticated, the limitations of these traditional methods have become more evident. This paper proposes a hybrid model framework combining Long Short-Term Memory (LSTM) networks and Feedforward Neural Networks (FNN), with the introduction of a Ridge Regression auxiliary model to reduce overfitting and improve classification performance through regularization. The proposed model has been experimentally evaluated on multiple publicly available Android malware datasets, such as Drebin and Android_Malware. Experimental results show that the LSTM + FNN + Ridge Regression model outperforms traditional machine learning methods, such as Support Vector Machines (SVM) and Random Forest, across multiple datasets. Moreover, when compared to the latest models, such as Transformer, the proposed model still demonstrates strong competitiveness. Through comparisons with traditional machine learning methods and Transformer-based models, it has been shown that the proposed model not only achieves better performance in terms of precision, recall, and other metrics, but also has an efficient training process and strong scalability.