Automated offense Prioritization for SIEM using ProbabilisticMachine Learning Models
Md Asif Khan, Akramul Azim, Farzaneh Abazari, Frank Eargle, Jeff Gardiner · 2024
Security Information and Event Management (SIEM) systems play a crucial role in cybersecurity by collecting, analyzing, and categorizing various events to detect and mitigate potential threats. These events are processed into offenses and then presented to Security Operations Center (SOC) analysts for further investigation and response. However, manual prioritization of offenses by SOC analysts can be challenging because of resource constraints and increasing volume of events. In our proposed solution for automating offense prioritization, we utilized the prediction probability and impact score derived from probabilistic machine learning (ML) models to address this challenge. The prediction probability indicates the likelihood of the occurrence of the offense, whereas the impact score signifies the severity of the potential consequences if the offense occurs. Additionally, we integrate time-based metrics such as Mean Time to Detect (MTTD) and Mean Time to Resolve (MTTR) into our offense prioritization framework. This combination of prediction probability, impact score, MTTD, and MTTR enables SOC analysts to prioritize offenses effectively, focusing their attention on those with high likelihood, high impact, and optimized response timelines. By addressing highly likely and high-impact offenses with efficient response processes, SOC analysts can optimize their response efforts, ensure the timely detection, and mitigate genuine security threats.