How to Train your Antivirus: RL-based Hardening through the Problem Space
Open Access
- 30 September 2024
- conference paper
- Published by Association for Computing Machinery (ACM)
- p. 130-146
Abstract
ML-based malware detection on dynamic analysis reports is vulnerable to both evasion and spurious correlations. In this work, we investigate a specific ML architecture employed in the pipeline of a widely-known commercial antivirus, with the goal to harden it against adversarial malware. Adversarial training, the most reliable defensive technique that can confer empirical robustness, is not applicable out of the box in this domain, for the principal reason that gradient-based perturbations rarely map back to feasible problem-space programs. We introduce a novel Reinforcement Learning approach for constructing adversarial examples, a constituent part of adversarially training a model against evasion. Our approach comes with multiple advantages. It performs modifications that are feasible in the problem-space, and only those; thus it circumvents the inverse mapping problem. It also makes it possible to provide theoretical guarantees on the robustness of the model against a well-defined set of adversarial capabilities. Our empirical exploration validates our theoretical insights, where we can consistently reach 0% Attack Success Rate after a few adversarial retraining iterations.Keywords
All Related Versions
Funding Information
- Research Fund KU Leuven (N/A)
- EPSRC Grant (EP/X015971/1)
- Horizon Europe KINAITICS project (01070176)
- Cybersecurity Research Program Flanders (N/A)
- Google ASPIRE research award (N/A)
This publication has 31 references indexed in Scilit:
- A Systematical and longitudinal study of evasive behaviors in windows malwareComputers & Security, 2022
- AntiViruses under the microscope: A hands-on perspectiveComputers & Security, 2022
- AIMED-RL: Exploring Adversarial Malware Examples with Reinforcement Learning2021
- Robust Malware Detection Models: Learning from Adversarial Attacks and DefensesForensic Science International: Digital Investigation, 2021
- Malware classification and composition analysis: A survey of recent developmentsJournal of Information Security and Applications, 2021
- Shortcut learning in deep neural networksNature Machine Intelligence, 2020
- An Adversarial Machine Learning Method Based on OpCode N-grams Feature in Malware Detection2020
- Transforming malicious code to ROP gadgets for antivirus evasionIET Information Security, 2019
- Adversarial Examples for Malware Detection2017
- Impeding behavior-based malware analysis via replacement attacks to malware specificationsJournal of Computer Virology and Hacking Techniques, 2016