Efficient Malware Investigation and Recognition Using Machine Learning Algorithms

Abstract

Malware is unique of the biggest problems that modern internet users have to deal with. Polymorphic malware is a new type of harmful software that is extra pliable than prior peers of bugs. Polymorphic malware continuously alters its signature characteristics in order to evade detection by conventional malware detection techniques. We applied various machine learning algorithms to detect malware or dangerous threats. A high detection ratio meant that the most accurate algorithm had been chosen to be used within the system. One advantage of the confusion matrix is its ability to track false positives and false negatives, providing deeper insights into the system’s performance. In particular, it revealed that machine learning algorithms like Naïve Bayes, Support Vector Machine (SVLM), Random Forest (RF), and K-Nearest Neighbor (kNN) can be used to detect harmful traffic on computer systems by calculation changes in correlation patterns. This approach enhances the effectiveness of malware detection and overall security in computer networks. The findings demonstrated that NB (87%), kNN (91.76%), SVM (92.41%), and RF (98.07%) performed well in terms of detection accuracy when compared to other classifiers. These findings are important as malicious software is growing more prevalent and sophisticated.