Generative Methods and Meta Learning for Cybersecurity
Abstract
Cyberspace is the digital communications network that that supports the internet of battlefield things (IoBT), the model by which defense-centric sensors, computers, actuators and humans are digitally connected. A secure IoBT infrastructure facilitates real time implementation of the observe, orient, decide, act (OODA) loop across distributed subsystems. Successful hacking efforts by cyber criminals and strategic adversaries suggest that cyber systems such as the IoBT are not secure. Three lines of effort demonstrate a path towards a more robust IoBT. First, a baseline data set of enterprise cyber network traffic was collected and modelled with generative methods allowing the generation of realistic, synthetic cyber data. Next adversarial examples of cyber packets were algorithmically crafted to fool network intrusion detection systems while maintaining packet functionality. Finally, a framework is presented that uses meta-learning to combine the predictive power of various weak models. This resulted in a meta-model that outperforms all baseline classifiers with respect to overall accuracy of packets, and adversarial example detection rate. The National Defense Strategy underscores cybersecurity as an imperative to defend the homeland and maintain a military advantage in the information age. This research provides both academic perspective and applied techniques to to further the cybersecurity posture of the Department of Defense into the information age.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 01, 2022
- Accession Number
- AD1181262
Entities
People
- Marc W. Chale
Organizations
- Air Force Institute of Technology