Do You Think You Can Hold Me? The Real Challenge of Problem-Space Evasion Attacks

Abstract

Android malware is a spreading disease in the virtual world. Anti-virus and detection systems continuously undergo patches and updates to defend against these threats. Most of the latest approaches in malware detection use Machine Learning (ML). Against the robustifying effort of detection systems, raise the evasion attacks, where an adversary changes its targeted samples so that they are misclassified as benign. This paper considers two kinds of evasion attacks: feature-space and problem-space. Feature-space attacks consider an adversary who manipulates ML features to evade the correct classification while minimizing or constraining the total manipulations. Problem-space attacks refer to evasion attacks that change the actual sample. Specifically, this paper analyzes the gap between these two types in the Android malware domain. The gap between the two types of evasion attacks is examined via the retraining process of classifiers using each one of the evasion attack types. The experiments show that the gap between these two types of retrained classifiers is dramatic and may increase to 96%. Retrained classifiers of feature-space evasion attacks have been found to be either less effective or completely ineffective against problem-space evasion attacks. Additionally, exploration of different problem-space evasion attacks shows that retraining of one problem-space evasion attack may be effective against other problem-space evasion attacks.

Problem-Space Evasion Attacks in the Android OS: a Survey

“…Second, exploring problem-space attacks is vital to test the realistic assumptions of feature-space evasion attacks. For example, Berger et al [36] showed that feature-space attacks in the Android domain do not serve as proxies for problem-space evasion attacks. In other words, feature-space attacks do not depict reality accurately.…”