Master of Science in Computer Science (MSCS)
Fabio Di Troia
Malware Detection, Adversarial Attacks, Machine Learning, Fake Malware Generation.
Malware detection is vital as it ensures that a computer is safe from any kind of malicious software that puts users at risk. Too many variants of these malicious software are being introduced everyday at increased speed. Thus, to guarantee security of computer systems, huge advancements in the field of malware detection are made and one such approach is to use machine learning for malware detection. Even though machine learning is very powerful, it is prone to adversarial attacks. In this project, we will try to apply adversarial attacks on malware detection models. To perform these attacks, fake samples that are generated using Generative Adversarial Networks (GAN) algorithm are used and these fake malware data along with the actual data is given to a machine learning model for malware detection. Here, we will also be experimenting with the percentage of fake malware samples to be considered and observe the behavior of the model according to the given input. The novelty of this project is given by the use of adversarial samples that are generated by the implementation of word embeddings produced by our generative algorithms.
Nagireddy, Vaishnavi, "Application of Adversarial Attacks on Malware Detection Models" (2023). Master's Projects. 1283.