Transforming Weakness into Strength: Improving Unreliable Malware Detection Methods

Abstract

This paper proposes a novel malware detection methodology that leverages unreliable Indicators of Compromise to enhance the identification of latent malware. The core contribution lies in introducing a sequence-based detection method that contextualizes unreliable IoCs to improve accuracy and reduce false positives. Unlike traditional methods reliant on predefined signatures or behavior analysis, this approach dynamically assesses system behaviors, focusing on suspicious actions and interaction patterns. Key contributions include a novel combination of unreliable IoCs with sequence alignment methods, an extensive mapping study of detection techniques, and initial experiments on a dataset of over 19,000 malware samples. Results demonstrate the method’s ability to cluster and identify malware families based on their behavioral signatures, even in its early developmental stage. This innovative approach shows promise for detecting previously unknown threats, establishing a foundation for advanced research in malware detection.

Keywords