University Data Can Fuel AI/ML Cybersecurity Research

Abstract

Research in AI/ML for cybersecurity requires high quality data showing both normal and abnormal events. Obtaining such data at meaningful scale has shown itself to be a challenge. Higher education institutions own large amounts of such data that they generate from their own IT and cybersecurity operations. This paper presents a vision for managed sharing of that data for research, acknowledging the challenges with implementing that vision and providing means for mitigating those challenges. A real-world example of such data sharing is described through a collaboration between a major, multi institutional security operations center, a AI/ML cybersecurity researcher, and an applied research center