Our ResearchIn Our Spare Time...
Aerstone is participating in the DHS “Protected Repository for the Defense of Infrastructure Against Cyber Threat” (PREDICT) initiative, which is designed to provide computer and network operational data for use in cybersecurity defensive R&D. PREDICT provides researchers and security development communities with real-world network operations data, to support testing and building next-generation cybersecurity solutions. Our primary research goal is to explore how insecure networks can be taught to protect themselves against common cybersecurity threats, by analyzing the nexus of machine-learning and cybersecurity.
We feel this initiative could improve security operations, by relieving the engineers from performing redundant tasks and allowing them to focus on mission-critical threat assessment activities. As a first step in this process, we are analyzing the dataset from the National Collegiate Cyber Defense Competition (NCCDC) held in 2012. The teams at the NCCDC were provided with a problem set to transform an insecure, fully functional commercial network into a more secure and maintainable network, and defend it against a live Red Team. By labeling all data elements as either “defender” or “attacker” actions, we plan to investigate whether Supervised Learning algorithms can predict the steps taken by the competing teams.