I followed an ‘alternative’ route into Higher Education, meaning that I had to be able to pick up new concepts and practices very quickly once I began my degree; I left the Army to pursue my passion for science. Previously , I served 5 years in HM Armed Forces. My time in the Army has given me the ability to work to a high standard of professionalism, especially when under pressure.
My interests lie in modelling the emergent behaviour of humans, this involves using reinforcement learning (RL) techniques to model the individual decision making process for performing actions and its effects on the emergent behaviour. In particular, combining elements from hierarchical, model based and multi-agent RL to look for new solutions/descriptions of observed phenomena.
|Complexity in the discrete non-linear Schrödinger Equation - This project focused on answering some key questions surrounding a new physical regime involving discrete nonlinear Schrödinger (dNLS)-type equations. The motivation was to explore, for the first time, the spontaneous pattern-forming instabilities in a system that describes the evolution of counter propagating light waves in an array of coupled nonlinear waveguides. The project involved a blend of theoretical analysis and computation, and built on earlier results involving the dNLS model applied in a ring-cavity context.|
|Modelling the best use of sleeping sickness diagnostics under
existing and emerging tools - By investigating combinations of diagnostic tools and developing ways to compare their efficiency and cost, we were able to identify key features of effective diagnostic algorithms. We go on to state the combinations that lead to the greatest prevalence reduction for the lowest price and identify some common characteristics among them. Our tests showed that diagnostic algorithms alone are not capable of reaching the WHO target - elimination of HAT by 2030. However, as part of a strategy including other interventions algorithms play a key role.
|Model Comparison for the Two Stage Decision Task - Presented here are several reinforcement learning models that propose to model sequential action choice behaviour, based on explanatory theory from neuroscience and psychology. Three recent models are focused on, each of which builds on the others. However, none of these models perform particularly well, with likelihoods of ∼ O(10 −80 ), neither do they outperform each other with any significance. There are a multiplicitude of factors that can contribute to their performance, which are explored from computational underpinnings, to form of the likelihood estimation function.|