The Mathematics of Cyber Defense
John A. Emanuello, Ahmad Ridley · Notices of the American Mathematical Society · 2022
The speed, complexity, and ubiquity of cyber-attacks has never been more apparent and the far-reaching impacts they have on society demonstrate the critical need for robust security solutions, which can reduce the success of cyber-attackers when (not if ) they compromise critical networks.Current cyber-defense capabilities are static and rules-based, i.e., they require a priori knowledge of the precise attacker tactics that will be employed.But this approach is unsustainable, given that malicious cyber actors rapidly change their approaches and chain their activities in complex and stealthy ways to thwart defenses.These challenges are driving a wide body of research and development of cybersecurity defensive solutions that are enhanced by artificial intelligence (AI), machine learning (ML), and data science.At their core, these approaches involve building mathematical models of cyber systems in order to derive information and devise strategies that enable their protection.However, unlike AI technologies applied in domains such as computer vision, natural language processing, and robotics, the complexities of the John A. Emanuello is a senior research mathematician at the National Security Agency (NSA).