A Robust Ensemble Learning Approach for Malware Detection and Classification

-

Abstract

In today's Internet world, many dangers threaten people's safety online every day. One big danger is harmful software called malware, like GoldenEyes, Heartbleed, Rootkit etc. This kind of software can make you lose important information or change it in a bad way. The usual ways of finding and stopping this software don't always work well. They take a lot of time and might not catch new kinds of harmful software. This paper introduces a robust ensemble approach for malware detection and classification. Leveraging a diverse and high-quality dataset, the proposed ensemble model combines three base classifiers Sequential model-1, 2, and 3 to enhance accuracy and resilience against evolving malware variants. Gray Wolf Optimization (GWO) is used to extract optimal features, optimizing model performance. Experimental results, obtained through rigorous comparative analysis with existing methods, demonstrate the superiority of the ensemble model, achieving a remarkable accuracy rate of 96.20%. This research contributes to the advancement of malware detection by offering a versatile and highly accurate solution capable of adapting to emerging threats, thereby bolstering cybersecurity efforts in an ever-evolving digital landscape.