Data Driven Skill Assessment for Cybersecurity Exercises

Measuring an individual’s cybersecurity competence in cybersecurity training is a challenging and complex task. It involves a wide range of skill assessments ranging from reconnaissance to post-exploitation in order to achieve some objectives on a target. Traditionally, flags (correct answer submissions) were used to measure individuals’ competence, which provides a holistic view of their performance in accomplishing a particular cybersecurity task. In this work, we investigate a deep learning-based diffusion model for measuring individual competence based on observed skills during a cybersecurity exercise. Our method helped identify weaknesses in an individual’s skill set, so a targeted intervention can be suggested to overcome the weaknesses and raise overall cybersecurity competence.