Honours and Masters project

Modern clinical decision-making is constrained by associative models that conflate correlation with causation and overlook interactions among patient factors. This project introduces a unified framework that fuses causal inference with graph neural networks to deliver interpretable, high-precision diagnosis. Using electronic health records, Double Machine Learning isolates causal drivers (e.g., treatment effects, biomarkers) from spurious associations while adjusting for confounders such as socioeconomic status.

Smart glasses that combine mixed reality head-mounted displays with computer vision and natural language understanding, such as the Apple VisionPro or Google XR Glass, have the potential to revolutionise the lives of people with low vision by providing access to information about their environment through augmented vision and audio.

This project is a multidisciplinary project between human-centred computing and data visualisation experts and water engineer experts in engineering and chemistry exploring new and immersive visual communication of complex ecosystems.

Access to cultural institutions, such as galleries and museums, is often compromised for people with disability. This includes people who are blind or have low vision (BLV). This project seeks to improve experiences within cultural institutions such as galleries and museums for BLV people, by applying AI and human-centred design principles to the creation of mediating artefacts and experiences.

Access to visual information, such as information graphics, is compromised for people who are blind or have low vision (BLV). Access is typically provided through written or verbal descriptions, or tactile graphics. These, however, are often provided by specialist producers which takes time and reduces the agency of the person for when they get the alternate format and also the ability to make their own interpretations.

🎯 Research Motivation

While many AI-powered coding assistants (e.g., GitHub Copilot, ChatGPT Code Interpreter) improve coding productivity, they are not optimized for pedagogical impact. CS students need not just code completion but understanding, feedback, and guidance that nurtures problem-solving and conceptual mastery.

Your research could bridge this gap by designing an AI IDE extension that acts as a mentor, dynamically adapting its feedback to the learner’s skill level, learning style, and progress.

We are interested in understanding genetic variation among individuals and how it relates to disease. To do this, we study genomic markers or variants called single nucleotide polymorphisms, or SNPs for short. A SNP is a single base position in DNA that varies among human individuals. The Human Genome Project has found that these single letter changes occur are all over the human genomes; each person has about 5M of them!  While most SNPs have no effect, some can influence traits or increase the risk of certain diseases.

Minimum Message Length (MML) is an elegant information-theoretic framework for statistical inference and model selection developed by Chris Wallace and colleagues. The fundamental insight of MML is that both parameter estimation and model selection can be interpreted as problems of data compression. The principle is simple: if we can compress data, we have learned something about its underlying structure.

Decision trees are powerful, interpretable models for prediction and classification that recursively partition the feature space into regions with homogeneous outcomes. Traditional decision tree algorithms like CART and C4.5 rely on heuristic splitting criteria and require ad-hoc pruning methods to prevent overfitting. In contrast, the Minimum Message Length (MML) framework provides a principled, information-theoretic approach to tree induction that naturally balances model complexity against data fit without requiring separate pruning phases.