A Robust Machine Learning Approach for Malware Detection in Linux

Abstract

This thesis has presented the implementation of machine learning algorithms for the detection of a class of ransomware that specifically targets Linux-based systems. Due to the increasing prevalence and complexity of such ransomware, traditional detection methods, which are primarily based on signature-based matching, are becoming obsolescent. This research endeavored to provide a fresh, novel, and effective way of combating malware that affects Linux operating systems in the form of a hybrid analysis novel methodology that combined static and dynamic analysis methods to extract maximum features from malware samples. The obtained features were then used to train four machine learning models, such as FNN, RF, LR and J48. The model’s effectiveness was verified using evaluation metrices such as accuracy, recall, F1-score, etc. The experimental results show the effectiveness of the applying machine learning techniques to improve malware detection and develop a robust means of enhancing cybersecurity in a world where threats are ever-growing