What is it about?
This paper describes a method that aids understanding of malware evolution and consequently improves detection of dangerous malware groups such as mutable malware through generative machine learning methods which involves generating malware samples and concurrently using them as training data to improve the detection ability of machine learning models.
Featured Image
Photo by Markus Spiske on Unsplash
Why is it important?
Our findings show that the generated adversarial samples are on average able to fool 63% of the AV engines tested on and the ML detectors are susceptible to the new mutants achieving an accuracy between 60%-77%. It is thus important to train ML models with the samples created to improve detection rates.
Perspectives
We enjoyed writing this paper and the findings were consistent with our previous work on the need for better detection models in defeating dangerous groups of malware such as metamorphic malware.
Kehinde Babaagba
Edinburgh Napier University
Read the Original
This page is a summary of: An Evolutionary based Generative Adversarial Network Inspired Approach to Defeating Metamorphic Malware, July 2023, ACM (Association for Computing Machinery),
DOI: 10.1145/3583133.3596362.
You can read the full text:
Contributors
The following have contributed to this page
