FEIT Research Project Database

Game theory for cyber defence

Project Leader: Tansu Alpcan
Collaborators: Dr Seyit Camtepe (Data61/CSIRO)
Sponsors: Data61/CSIRO
Primary Contact: Tansu Alpcan (tansu.alpcan@unimelb.edu.au)
Keywords: cybersecurity; defence; game theory; Internet of Things (IoT); machine learning; optimisation
Disciplines: Computing and Information Systems,Electrical & Electronic Engineering

Machine learning (ML)-driven data analytics have been increasingly used to identify malicious threats to critical infrastructure and to make cybersecurity decisions, eg: artificial intelligence (AI)-based advanced threat detection/prevention solutions deployed in IoT, Cloud, healthcare, and smart grid systems. However, complex nature of ML/AI solutions makes it very hard to verify, validate and adapt these decisions under maliciously tailored data, compromised models, faulty algorithms, implementation flaws, system dependencies and human (ie: attacker, defender and victim) choices. Game theory provides a powerful tool that can be used to model, predict, and evaluate these malicious behaviours. This project aims to investigate game-theoretic approaches to cyber-defence, especially in conjunction with ML/AI methods and critical system applications.

The interested candidate should have a solid background in (convex) optimisation and machine learning with a willingness to learn new theories and practical application areas. The project involves collaboration with a prominent researcher from Data61/CSIRO, Dr Seyit Camtepe, and associated sholarship and internship opportunities.