There is a gap in the Cyber Security employment market for people combining Mathematics and Data Science skills, for which this course was specifically designed. Data Science Toolbox and Anomaly Detection are two taught courses that together leave students with very transferrable Data Science skills that are in especially high demand, covering core Machine Learning concepts including:
Exploratory Data Analysis, Statistical Testing and cross-validation, Latent Embedding, Non-parametrics, Classification, Random Forests, Boosting, Topic Models, Understanding Algorithms, Neural Networks, Parallel Algorithms, Spark and massively parallel data, and Ethics and Privacy.
The Data Science skills developed in this course are highly transferrable, and developed with practical examples. Combining these courses with Complex Networks, students are equipped for any Data Science career.
The Project Preparation and Project will leave all students with a unique set of skills that involve practical experience on real-world cyber security datasets.
The Introduction to Mathematical Cybersecurity unit allows mathematics students to quickly get up to speed on Cyber Security details.
Students may also be interested in the Engineering Data Science MSc.
Data Science is at the centre of a revolution in Cyber Security. We all run anti-virus software, but this is not enough to stop a determined attacker, who will always get in (for example, by phishing). Mathematical Cyber Security is used to detect attackers after they get into your network, before they can do damage. This figure shows a data science approach to detecting a (simulated) attack. The intruder hacks computer A, and uses this foothold to move to B, C and finally D where your critical information is held. They then move that data back through C and B to bring it home. This can be detected by (1) recording data about “sessions”, and (2) spotting connections that overlap suspiciously. Then (3) sophisticated statistical tools can confirm the attack and trigger a network shutdown.
Some cyber-security research undertaken by the Institute for Statistical Sciences includes:
- ‘Detecting weak dependence in computer network traffic patterns by using higher criticism’, Price-Williams, M, Heard, N & Rubin-Delanchy, P, 2019. Journal of the Royal Statistical Society. Series C: Applied Statistics, vol 68., pp. 641-655
- ‘Anomaly detection for cyber security applications’, Rubin-Delanchy, P, Lawson, DJ & Heard, NA, 2016. in: Dynamic Networks and Cyber-Security. World Scientific Publishing Co., pp. 137-156
- ‘Disassortativity of computer networks’ Rubin-Delanchy, P, Adams, N & Heard, N, 2016.
- ‘Network-wide anomaly detection via the Dirichlet process’ Heard, N & Rubin-Delanchy, P, 2016.
- An approximate framework for flexible network flow screening Lawson, D. J. & Adams, N. M., 26 Sep 2014.
- Statistical frameworks for detecting tunnelling in cyber defence using big data Lawson, D. J., Rubin-Delanchy, P. T. G., Heard, N. A. & Adams, N. M., 26 Sep 2014, p. 248-251.
- Three statistical approaches to sessionizing network flow data Rubin-Delanchy, P. T. G., Lawson, D. J., Turcotte, M. J., Heard, N. A. & Adams, N. M., 26 Sep 2014, p. 244-247.
