A Novel IChOA–CNN-LSTM Model for Android Malware Detection Using Opcode-Based Feature Selection and Optimization

Abstract

The rapid growth of Android applications has resulted in an increase in security concerns, specifically malware attacks. Traditional signature-based and heuristic detection technologies fail to keep up with the changing nature of malware. To address this issue, deep learning-based algorithms provide a viable solution by utilizing sophisticated feature extraction and classification techniques to improve detection accuracy. This paper presented an Improved Chimp Optimization Algorithm-based CNN-LSTM (IChOA-CNN-LSTM) technique for detecting Android malware. The procedure starts with an Android malware dataset, which goes through data pre-processing to clean and modify it for best analysis. To extract significant characteristics, a feature selection procedure is used in conjunction with an opcode-based model. Furthermore, the IChOA-CNN-LSTM technique uses the IChOA-CNN-LSTM methodology to improve malware detection. The model improves feature selection by combining an enhanced transformer with an RNN model and a softmax function for better classification. Finally, the Snake Optimizer Algorithm (SOA) is used to fine-tune parameters for the best detection performance. Extensive experimental findings show that the IChOA-CNN-LSTM technique is successful for detecting Android malware. The system's performance is measured using key measures like accuracy, precision, recall, and F-score in addition to a strong malware detection architecture.