AI-enabled approach for enhancing obfuscated malware detection: a hybrid ensemble learning with combined feature selection techniques

Abstract

In an era where the relentless evolution of cyber threats necessitates the perpetual advancement of security measures, the detection of obfuscated malware has emerged as a formidable challenge. The clandestine tactics employed by malicious actors demand innovative solutions that transcend conventional approaches. In this context, this research present a groundbreaking research endeavor that redefines the frontiers of obfuscated malware detection using artificial intelligence. In this research, a comprehensive methodology is introduced that combines three pivotal feature selection techniques: correlation analysis, mutual information, and principal component analysis. This hybrid approach not only enhances the discrimination of meaningful features but also ensures the efficiency and effectiveness of the feature subset, thus mitigating the curse of dimensionality. To harness the full potential of these meticulously selected features, an array of ensemble-based machine learning algorithms, including AdaBoost, stacking, random forest, bagging, and voting, is deployed. Amongst these, our findings demonstrate that AdaBoost emerges as the preeminent choice, achieving unprecedented levels of performance. The outcomes underscore the profound impact of our research in the realm of obfuscated malware detection, a paradigm shift that reimagines the very essence of security. In a world where cybersecurity challenges continually escalate, our research represents a pivotal milestone in the unceasing battle to safeguard digital landscapes. It is an exultant testament to the boundless potential of innovative feature selection techniques and the supremacy of AdaBoost within the domain of malware detection.