Motivated by seeking solutions that are both explainable and scalable for high-dimensional regression, classification, and clustering tasks, this thesis develops a univariate feature selection method, a probabilistic extension for the Self-Organising Map clustering algorithm, the development of two sparse Bayesian regression models for improved efficiency with a Laplace-approximated generalisation for non-Gaussian likelihoods, and a Python package to process and model high-dimensional microbiological marker datasets.
On Sparse Solutions and Topology Learning for High Dimensional Problems in Microbiology
Taylor, J. (Author). 25 Jun 2025
Student thesis: Doctoral Thesis › PhD