Optimizing Adversarial Attacks on ML-Powered Malware Detection Systems

Malipeddi, Vikas Varma (2024) Optimizing Adversarial Attacks on ML-Powered Malware Detection Systems. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (1MB) | Preview

Abstract

This report demonstrates a new approach to the development of optimizing adversarial attacks on ML-powered malware detection systems. Contradictory to the existing methodologies that accepts unlimited access and the queries to the target detection system, our research project handles the realistic limitations faced by adversaries in the actual cybersecurity conditions. In these scenarios, the attackers can usually encounter limited access to the detection system and have a restricted number of queries. The main objective is to discover and implement the adversarial method techniques that not only potentially escape the machine learning-based malware detectors but also handle within the boundary of a inhibited query budget. The study focuses on the enhancing our understanding towards the limitations and the vulnerabilities including in current based machine learning malware detectors within the real-world cybersecurity topic. By connecting a practical viewpoint, our research goals to contribute to the development of more robust defense mechanisms. An innovative implementation includes the utilization of the surrogate model to generate the adversarial malware samples, which leads to leveraging the conception of transferability. This approach put forwards that the successful attacks on the surrogate model can carry over and effectively compromise the target model. Through this research, we seek to offer the important insights into the constantly evolving environment of adversarial attacks on machine learning based malware identifying techniques and uplift to the development of adaptable defense methodologies in the cybersecurity domain.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Trinh, Anh Duong UNSPECIFIED
Uncontrolled Keywords:	Malware detector; Machine learning; Cybersecurity
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science Q Science > QA Mathematics > Computer software > Computer Security T Technology > T Technology (General) > Information Technology > Computer software > Computer Security Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Artificial Intelligence
Depositing User:	Ciara O'Brien
Date Deposited:	30 May 2025 14:17
Last Modified:	30 May 2025 14:17
URI:	https://norma.ncirl.ie/id/eprint/7716

Actions (login required)

View Item