Ensemble Machine Learning and Feature Selection for Effective Malware Detection

Abstract

Malware detection is a crucial aspect of cyber security, as malicious software leads to a significant threat to the integrity and security of digital systems. With the constantly evolving nature of malware, traditional signature-based detection methods struggle to keep pace with appearing threats. In recent years, machine learning has become visible as a promising solution to enhance malware detection capabilities, leveraging its ability to identify complex patterns and adapt to new attack vectors. This paper presents an extensive study of malware detection using machine learning techniques. Machine learning algorithms, including adaboost ensemble learning, stacking ensemble learning, hard voting ensemble learning, and soft voting ensemble that have been employed to tackle the challenge of malware classification. Furthermore, we explore the Variance threshold and wrapper based forward feature engineering process, delving into the extraction of relevant features from malware samples to enable effective machine learning-based detection. In this paper, Malware detection techniques applied on the dataset CCCS-CIC-AndMal-2020 and get the accuracy of 99.48%.