I enjoy making things. Here are a selection of projects that I have worked on over the years.
This report explores using machine learning to classify age and gender from racket kinematics during table swings, using a modified dataset from DRYAD. It compares KNN, SVM, and One-vs-Rest Logistic Regression, revealing how racket motion features influence classification performance.
PHAML enhances machine learning by integrating topological features from persistent homology with traditional data inputs, capturing global structure through Vietoris–Rips filtrations. This approach improves classification accuracy and robustness compared to baseline models using only conventional features.
FLAN-T5 fine-tuning and LoRA for biomedical lay summarisation. Full implementation on GitHub.
