Optimizing Malware Detection Through Advanced Ensemble Methods

One of the biggest problems that institutions & organizations face is information security. The frequency and scope of cybercrime have increased recently, as new methods for stealing, altering, and destroying data or taking down computer networks are emerging daily. Malware is one kind of intrusion that can occur into information systems that handle sensitive data. When malware is introduced into a computer system, the attacker gains full or partial access to the system’s vital data. An ensemble classification-based detection of malware algorithm is proposed in this paper. A layered ensemble of dense (fully connected) and CNN handles the initial step of classification, and a meta-learner handles the last stage of classification. We investigate and contrast 14 classifiers for a meta-learner. Eight ML techniques are KNN, SVM, Random Forest, AdaBoost, Decision Tree, Logistic Regression, Gradient Boosting, XGB, and the suggested approach are utilized as a baseline comparison. We report on the studies conducted using the CSV or XLS file datasets for the classification of malware. The results all demonstrate how simple it is to identify malware in software files using the suggested method.