AMDF: An Additive Malware Detection Framework for Knowledge Enhancement Using Heuristic Approaches

Abstract

This research paper presents an innovative framework for malware detection that leverages genetic algorithms as a heuristic approach to enhance knowledge acquisition and decision-making capabilities. Traditional signature-based detection methods face increasing challenges in identifying polymorphic and metamorphic malware that continuously evolve to evade detection. The proposed Additive Malware Detection Framework (AMDF) incorporates genetic algorithms to adaptively learn from new malware variants and optimize detection parameters over time. Through experimental validation using a diverse dataset of 5,000 malware samples across five distinct families, the framework demonstrated a 94.7% detection accuracy, outperforming conventional approaches by an average of 17.3%. Additionally, the framework exhibited resilience against zero-day attacks with a 78.2% detection rate for previously unseen malware variants. The knowledge enhancement component of the framework facilitates continuous learning, reducing false positives by 34.8% compared to static detection methods. This research contributes to the cybersecurity domain by establishing an adaptive and evolutionary approach to malware detection that addresses the limitations of traditional methodologies while providing robust protection against emerging threats.