Performance Analysis of Machine Learning Classifiers for Malware Detection

In today’s cybersecurity landscape, the continuous evolution and diversification of malware present a significant threat, necessitating its effective detection and classification. With malware types becoming more sophisticated, the demand for advanced detection techniques is paramount. The detection and classification of malware constitutes a critical challenge, necessitating the deployment of sophisticated machine learning (ML) techniques. This study proposes a methodology that leverages a variety of machine learning (ML) models, including Naive Bayes, Decision Trees, Extra Trees, Random Forest, AdaBoost, Gradient Boost, LightGBM, CatBoost, and Support Vector Machines, fine-tuned for optimal performance. Conducting a comprehensive performance analysis across binary, 4-class, and 16-class classification tasks on the extensive CIC-Malmem2022 dataset, this research aims to identify the most effective classifier for malware detection and classification. This study emphasizes LightGBM’s superiority over other classifiers due to its strong performance with categorical features and large datasets. This highlights its potential to bolster cybersecurity by enhancing malware detection and optimizing ML classifiers against evolving threats.