Malware Detection Using Machine Learning

Abstract

Malware is any software or a computer program that performs malicious actions on a legitimate user’s computer system such as information stealing and spying. While the number of malware attacks is rapidly increasing, the major task of cyber security is to protect computer systems from malware attacks which can be done through efficient malware detection. Currently used signature-based methods for malware detection do not provide accurate results in the case of polymorphism or zero-day attacks. Hence, this paper focuses on detecting malware using machine learning techniques. A program can be trained to identify if certain software is malicious or not. By using a Python script, we train a classifier such that it can detect whether Portable Executable (PE) format files are malicious or non-malicious. Five different classification algorithms – Gaussian Naive Bayes, AdaBoost, Gradient boosting, Decision tree, Random Forest classifiers are applied and the best classifier is chosen for prediction by comparing their results in terms of accuracy. The overall best performance is expected to be given by Random Forest classifier with accuracy above 95%.