Robust/adversarial learning

We are also interested in how to reduce the side effect of noise on the instance, which may be caused by the failure of sensors or even malicious attacks. We human have the ability to correctly recognise the objects even there are noise (e.g., we can easily recognise human faces under extreme illumination conditions, when …

Statistical (deep) learning theory

Deep learning algorithms have given exciting performances, e.g., painting pictures, beating Go champions, and autonomously driving cars, among others, showing that they have very good generalisation abilities (small differences between training and test errors). These empirical achievements have astounded yet confounded their human creators. Why do deep learning algorithms generalise so well on unseen data? …

Transfer learning

Just like human, machine can also find the common knowledge between tasks and transfer the knowledge from one task to another one. In machine learning, we can exploit training examples drawn from some related tasks (source domains) to improve the performance on the target task (target domain). This relates two terms in machine learning, i.e., …

Leading digital infrastructure

Underpinning our leading AI research, we will expand on the AIS platform, designing and implementing a suite of digital tools to accelerate AI imaging research, better utilise the university’s hardware investments, securely collaborate with health and commercial partners, and provide a provenance trail for establishing trust.

Explainable AI

In this project, we will study the research area “explainable AI” that enables the outputs from the AI systems to exhibit a certain degree of explainable capability in order for clinical experts to better understand how AI systems diagnose diseases. The feedback from the clinical experts could in turn help us to develop better deep …

Imaging enhancement using AI

This project aims to exploit advanced AI technologies to enhance the quality of MRI data to make it comparable to that obtained from time consuming high-quality research protocols. The increased image data quality opens up the possibility of reliably applying advanced analysis methods to standard clinical data. The methods will be demonstrated in a number …

Imaging biomarkers from AI federated learning

This MRFF project seeks to build a novel, hybrid AI learning ecosystem to generate clinically-relevant biomarkers of disease progression for the common, disabling neurological condition, multiple sclerosis (MS).  The MSBASE-XNAT imaging repository and I-MED clinical radiology site data will respectively form the key components of a unique central-federated AI learning environment, yielding algorithms that, validated …

Human intelligence in AI loop

In this project, we will investigate how to introduce human intelligence in the loop when developing advanced AI methods, to progressively improve disease diagnosis results by taking advantage of both AI and human intelligence.

Imaging analysis and AI

In this project, we will focus on cohesively developing advanced AI technologies to revolutionise imaging-based diagnosis of disease and extend our work towards commercial applications and clinical deployment. Specifically, we will first target three disease area cohorts where AI can greatly assist medical professionals: cancer (breast, brain, melanoma), lung diseases (e.g., COVID-19), and neurological conditions.


Prof Stefan B. Williams – Director, Digital Sciences Initiative
Faculty of Engineering, University of Sydney NSW 2006 Australia
+61 2 9351 8152