Exploring malware complexities: a behavior and characteristic analysis approach for robust and accurate cybersecurity

Android malware detection has emerged as a recent hot problem with the rapid advancement and proliferation of telecommunication networks. Besides, malware is envisioned as a catastrophic form of cybercrime that requires immediate action and timely detection to prevent unauthorized access to networks. Traditional methods for Android malware detection and categorization face several challenges such as poor detection accuracy, scalability issues, and require more computational resources. Therefore, this research work leverages ensemble support vector machine-based crossover equilibrium (ESVM-CE) for malware detection and categorization to better streamline these challenges of predominant concern. The devised model employs visual-based malware samples representing diverse malware families for understanding the complex characteristics that separate malware from benign files. The ESVM-CE model initially performs malware image sample preprocessing to ameliorate the data quality, making it appropriate for malware detection. By leveraging a convolutional neural network (CNN), the complex patterns inhered in the preprocessed raw data are learned in an easier and computationally efficient manner due to its simpler architecture. Moreover, the derived malware characteristics are then provided as input to the Ensemble SVM optimized by crossover-based equilibrium optimizer (CE) for malware threats detection and categorization. The ESVM-CE method has been comprehensively evaluated with publicly accessible datasets including, the Malimg dataset, BIG 2015 dataset, and Male Vis dataset for diverse evaluation measures, and compared against baseline data-driven approaches. By employing those datasets, the evaluation results show that the ESVM-CE method is more robust than baseline methodologies by achieving 97.54% precision, 98.52% recall, and 98.94% accuracy. Overall, the research work contributes to the enhancement of system reliability and security by performing the detection and categorization of Android malware threats accurately and efficiently.