In the real world, decisions are typically made under some level of uncertainty as we rarely have access to perfect information or complete knowledge about a problem. Humans learn to be more cautious under ambiguous information, but in many cases we struggle to teach machines to behave in the same way. AI systems are now ubiquitous in society and are becoming ever more responsible for decisions that impact human safety. Examples of AI making life critical decisions include medical imaging diagnosis systems and self-driving cars. To ensure the safety of these decisions, it is essential that these models can be imbued with a useful degree of uncertainty.

The objectives of this project are to advance our understanding of how to effectively and efficiently produce and use uncertainty estimates from AI models in downstream medical image analysis and computer vision tasks. This project will leverage modern deep learning methods to develop novel probabilistic AI models, which are better able to understand the inherent uncertainty in the problems they are tasked with. It will also investigate how this information can be incorporated into decision making. The specific application area is flexible, depending on the student’s interest. Some potential models of interest include image registration and tracking, segmentation, quantitative MR image analysis, image translation and depth estimation.

This is a computational project, which would suit a student with good mathematical and programming skills and a keen interest in building robust computer vision models. Eligible applicants will have recently received an MSc and/or a First or high 2:1 BSc in a relevant subject.

This project is with Dr Ivor Simpson, a recently appointed lecturer in the Predictive Analytics Lab at the University of Sussex, which is situated on the outskirts of Brighton adjacent to the South Downs national park.

Informal enquires are welcome, please email i.simpson@sussex.ac.uk.