Statistics in Cyber-Security - Dr Nick Heard, Imperial College London

Explore statistical approaches for enhancing cyber-security defenses in this comprehensive lecture by Dr. Nick Heard from Imperial College London. Delve into the application of data science techniques for identifying and preventing cyber-attacks and network misuse within enterprise computer networks. Examine various statistical and probability model-based methods for analyzing cyber data, ranging from micro-level models of activity on individual graph edges to representations of full network graphs. Learn about different data sources available in enterprise networks, various levels of resolution in cyber-security analysis, and specific techniques such as automated edge detection, Bayesian mixture models, and intensity-based models. Discover related applications like topic modeling and Bayesian change detection, and gain insights into the importance of robustness and dynamic feature formation in cyber-security analytics.

Intro
Collaborators
Statistics in cybersecurity
Sources of data
Different levels of resolution
Edges
Automated edges
G test
Bayesian mixture model
Human generated events
Edge scoring
New edges
Intensitybased models
Example
Related Applications
Topic Modelling
Notation
Robustness
Bayesian inference
Weighted moving averages
Dynamic feature formation
Bayesian change detection
Summary