RSS feed Add to your Facebook page Watch us on Youtube

Activity title

Deep Machine Learning for Cyber Defense

Activity Reference

IST-163 (IWA)

Panel

Information Systems Technology

Security Classification

PUBLIC RELEASE

Status

Active

Activity type

RTG

Start date

2018

End date

2021

Keywords

Artificial Intelligence, Deep Learning, Feedbackbased Learning, Knowledge based learning, Machine Learning, Rulebased Learning

Background

Cyber threats are more advanced and strategic and can manifest from anywhere in the World. Terrorism in cyberspace is at an all-time high and NATO countries have to enhance their strategic landscape by using 21st century technology to mitigate cyber threats to military systems, platforms and mission. Deep machine learning (ML) is a state of the art technology that the military can implement to enhance its strategic cyber position and create a defense that not only addresses threats of today, but threats 10 and 20 years into the future. The main goal of the proposed activity is to consolidate the NATO-wide knowledge in the field of deep ML and cyber defense, identify the gaps between civilian solutions and military needs, and collaborate with other NATO countries to use data processing, share data and pursue the transfer of the most promising technologies and applications to the military domain. Special attention will be paid to the alignment of terminology with related activities within other NATO initiatives, particularly the NATO Modelling and Simulation Group. As such, it will address a multidisciplinary audience from the fields of artificial intelligence, machine learning, modelling & simulation, and systems engineering. The working group efforts will focus on machine learning encompassing the deep learning aspect.

Objectives

Contribute to the improved understanding of how to apply Deep Learning (machine learning) to cyber defense. Deep Learning revolves around a feedback-based machine learning mechanism and the working group will examine main areas of interest: • Holistic gaps to bringing deep machine learning into cyber operations o Capability Development: understanding the capabilities that deep machine learning in cyber defense provides. o Techniques for learning through collecting, processing and understanding data: Testing, Evaluation, Verification, and Validation. • Technical Gaps o Environmental Perception – specifically related to military domains and missions. Better Situational Awareness. o Shared machine learning across (national) cyber tools: machine learning teaming, multi-agent teaming. o Interoperability with other cyber defense tools: the applicability of open source tools for military. Not a duplication of another NATO IST working group. o Shared data sets: collection and sharing of labeled datasets for improved performance from currently existing machine learning tools, and for research into new machine learning and AI tools for cyber defense.

Topics

• Improving, tailoring or optimizing existing Deep ML techniques for cyber defense applications o Architectural design to enable Machine Learning in simulations. o Critical factors and potential barriers for application of Machine Leaning. • Data collections for AI and ML o Deep ML Joint experimentation with varying levels of deep machine learning on militarily relevant missions. o Collect data from many varied platforms that can be shared by participating nations for use in developing improved AI and ML algorithms. o Experiment in varied environments in participating countries. • Resilient and adaptive behaviors

Contact Panel Office