Burning the Adversarial Bridges: Robust Windows Malware Detection Against Binary-level Mutations
This addresses the problem of robust malware detection for cybersecurity practitioners against adversarial attacks, representing an incremental improvement with specific preprocessing and representation methods.
The paper tackles the problem of adversarial malware detection by identifying volatile features in software binaries that make detection systems vulnerable to binary-level mutations, and proposes software preprocessing steps and a graph-based section-dependent information extraction scheme to reduce this attack surface. Their approach achieves 88.32% AUC for malware detection and maintains 88.19% AUC under combined binary manipulation attacks.
Toward robust malware detection, we explore the attack surface of existing malware detection systems. We conduct root-cause analyses of the practical binary-level black-box adversarial malware examples. Additionally, we uncover the sensitivity of volatile features within the detection engines and exhibit their exploitability. Highlighting volatile information channels within the software, we introduce three software pre-processing steps to eliminate the attack surface, namely, padding removal, software stripping, and inter-section information resetting. Further, to counter the emerging section injection attacks, we propose a graph-based section-dependent information extraction scheme for software representation. The proposed scheme leverages aggregated information within various sections in the software to enable robust malware detection and mitigate adversarial settings. Our experimental results show that traditional malware detection models are ineffective against adversarial threats. However, the attack surface can be largely reduced by eliminating the volatile information. Therefore, we propose simple-yet-effective methods to mitigate the impacts of binary manipulation attacks. Overall, our graph-based malware detection scheme can accurately detect malware with an area under the curve score of 88.32\% and a score of 88.19% under a combination of binary manipulation attacks, exhibiting the efficiency of our proposed scheme.