AI Security and Privacy
AI technologies including Machine Learning and Deep Learning technologies have been applied to a variety of different application domains such computer vision, image processing, audio data processing, networking, and cyber security. Our research group has applied various AI techniques to the above-mentioned areas. There are a lot of emerging cyber security and privacy issues for AI based systems and our group has been doing research on how to provide security, privacy and explainability of AI systems against various security and privacy threats. Our research topics are summarised as follows:
Cyber security using AI technologies
Adversarial Machine Learning
Privacy for Machine Learning and Deep Learning algorithms and techniques
Researchers: Dr Guangdong Bai, Dr Siamak Layeghy, A/Prof Dan Kim, Prof Ryan Ko, Dr Abigail Koay, A/Prof Marius Portmann, Dr Guowei Yang.
Current Projects
Resilient Learning-based Defense under Adversarial and Uncertain Environments (funded by the US Army International Technology Center-Pacific (ITC-PAC), 2020–2023)
This project has three main objectives: 1) develop resilient intrusion detection models for enterprise networks, software defined networks, and in-vehicle networks under uncertain environments; 2) develop a novel moving target defence against AI powered attacks; 3) intrusion response system under uncertainly.
Machine Learning for Automated Network Anomaly Detection, Cyber Security and Analysis (2019-present)
Researchers: Marius Portmann
Privacy preserving federated deep learning for medical imaging (UQ Cyber Seed Funding, 2021-2022)
Researchers: Dr Guangdong Bai
Automated Fuzzing for Deep Learning systems
Researchers: Guowei Yang