Honours and Masters project

The last two decades have witnessed a sharp rise in the amount of data available to business, government and science. Data visualisations play a crucial role in exploring and understanding this data. They provide an initial grasp of the data and allow the assessment of findings of data analytics techniques. This reliance on visualisations creates a severe accessibility issue for blind people (by whom we mean people who cannot use graphics even when magnified).

Despite an increase in the usage of AI models in various domains, the reasoning behind the decisions of complex models may remain unclear to the end-user. Understanding why a model entails specific conclusions is crucial in many domains. A natural example of this need for explainability can be drawn from the use of a medical diagnostic system, where it combines patient history, symptoms and test results in a sophisticated way, estimate the probability that a patient has cancer, and give probabilistic prognoses for different treatment options.

Using artificial intelligence software and unique algorithms for predictive analytics that incorporate modelling, machine learning, and data mining, we analyse, model, and build an individual’s baseline health profile against thousands (eventually millions) of similar people and their data points, along with decades of evidence-based medical and population research. Our previous work focused on the prediction of Diabetes Type Two – a major debilitating chronic disease, and a significant contributor to global deaths.

This project focuses on simulating the organic growth of criminal communication networks by leveraging techniques such as Reinforcement Learning and Graph Theory. The goal is to curate a synthetic dataset that models the evolving structure and dynamics of illegal networks, taking into account factors like social connections, communication patterns, and resource allocation. By using graph-based models, the project aims to create realistic representations of how criminal groups form, expand, and operate under various conditions.

This project focuses on developing a gait detection system leveraging computer vision techniques to recognize individuals in security footage, even when their faces and skin are obscured. Criminals often cover their faces to avoid recognition, making traditional facial recognition unreliable. By analyzing key features of an individual’s gait, such as walking style, speed, and direction, this system provides a robust alternative for visual attribution.

This project aims to develop a computer vision system capable of detecting and classifying domestic geographic landmarks in images and video content. By categorizing locations such as “childcare centre”, “school”, “shopping centre”, “hotel”, “bus stop”, “train station”, and so on, the system provides a reliable way to identify key landmarks in urban and suburban environments. Using advanced machine learning techniques, the model processes visual data to recognize distinct architectural features, signage, and contextual cues associated with each landmark type.

As cities face unprecedented growth, the need for tools that can integrate diverse knowledge sources ranging from geospatial data to the nuanced intuition of urban planners is critical. This research will explore how Bayesian Networks can be adapted to serve as configurable, transparent models that empower decision-makers to weigh alternatives involving complex factors such as development yield, urban zoning, locality to services and infrastructure capacity.

Communication networks show interaction between people over time, and are key to the identification of criminal networks and criminal activity. This project will investigate how future events might be able to be predicted, based on dynamic graphs representing prior interpersonal communications. The project will consider (a) how Graph Neural Networks can best be used for this machine learning task; (b) how visualisation techniques can best depict both known-past and predicted-future events.

Cyber Threat Intelligence (CTI) plays a vital role in today's cybersecurity landscape by collecting and analysing data about current and potential threats, providing insights to better understand, mitigate and respond in this ever-evolving environment. A core component of CTI is the identification of adversarial Tactics, Techniques, and Procedures (TTPs), which describe how attackers operate at a strategic and operational level.

This project focuses on understanding and applying control charts as a tool for monitoring and improving process quality. It involves designing some basic control charts and evaluating their performance in detecting process variations under different conditions. The evaluation will be based on key metrics such as Average Run Length (ARL), false alarm rate, and detection speed, providing insights into the effectiveness of various chart types in maintaining quality standards.