Supervisors

Suitable for

Abstract

Ransomware attacks are increasing rapidly every year. While signature-based malware detection methods work well for detecting and stopping known malware, attackers can bypass the detection using obfuscation techniques or zero-day attacks. There is therefore a need for better detection mechanisms that are able to predict new forms of malware and react against them.

This project aims at exploring malware detection to develop a better understanding of the differences that make malware different from normal processes or files. It will further seek to implement a machine learning (ML) tool that would help in detecting malicious behaviour efficiently, so that malware infection and propagation can be stopped. The ML classifiers used will depend on the malware family explored as well as data available.