Hybrid Optimization Based Long Short-Term Memory for Android Malware Detection

Malware detection is process of detecting the suspicious files and activities, that can classify as benign and malicious depends on different characters like capability of infection, structure of malware and speed rate. In this paper, proposed a Particle Swarm Optimization and Ant Colony Optimization (HPSALO) based Long Short-Term Memory (LSTM) model for malware detection and classification. The combination of these two algorithms that leads to the balance between exploration and exploitation. The dataset utilized for the research is DREBIN dataset and process of data cleaning is performed by NumPy library. The feature selection is performed by combination of hybrid PSALO algorithms and selected features are given to Long Short-Term Memory (LSTM) classifier that classifies as benign and malware files. The performance of proposed method is estimated by performance metrics of accuracy, precision, recall and f1-score. The proposed method attained accuracy of 99.38%, precision of 99.30%, recall of 99.08% and f1-score of 99.42% which is comparatively higher than other existing methods like Whale Optimization Algorithm-LSTM (WOA-LSTM), Cat Swarm Optimization-LSTM (CSO-LSTM), Particle Swarm Optimization-LSTM and Ant Lion Optimization-LSTM.