Theodore Kypraios

Leverhulme Trust Research Fellowship

After several years balancing research, teaching and administration responsibilities — including a recent spell as Head of the Statistics and Probability Section in the School of Mathematical Sciences at the University of Nottingham — I’m incredibly excited to have the chance to focus deeply on research again.

A Research Fellowship will give me the time and space to pursue new ideas in the topical area of sample-based-inference (SBI) or stochastic processes, collaborate with people, travel and, most importantly, learn new things!

I’m very grateful to The Leverhulme Trust for this opportunity — and looking forward to what comes next!

Recent paper published in Bayesian Analysis

Epidemics are inherently stochastic, and stochastic models provide an appropriate way to describe and analyse such phenomena. Given temporal incidence data consisting of, for example, the number of new infections or removals in a given time window, a continuous-time discrete-valued Markov process provides a natural description of the dynamics of each model component, typically taken to be the number of susceptible, exposed, infected or removed individuals.

Fitting the SEIR model to time-course data is a challenging problem due incomplete observations and, consequently, the intractability of the observed-data likelihood. Whilst sampling based inference schemes such as Markov chain Monte Carlo are routinely applied, their computational cost typically restricts analysis to data sets of no more than a few thousand infective cases. Instead, we have developed a sequential inference scheme that makes use of a computationally cheap approximation of the most natural Markov process model.

This work has been accepted for publication in a Bayesian Analysis is part of Sam Whitaker supervised by Andrew Goligthly (Durham) and Colin Gillespie (Newcastle) and be accesed from this link.

Identifying sperm whales and other cetaceans by their flank or fluke!

Recent work led by PhD student Sammie Fuller (supervised by Mike Pound, Silvia Maggi, Barbara Mussi and myself) develops machine learning tools to automate the identification of individual Mediterranean sperm whales from photographs—a crucial step for conservation of this endangered species. Traditionally, whale research relies on labor-intensive manual matching of fluke images. We extend this by showing that flank images can also be used effectively, training deep learning models that achieved over 80% accuracy on flukes and nearly 77% on flanks. We also provide a practical interface to support researchers in identifying known whales and adding new individuals. These methods are adaptable to other cetaceans, offering scalable, non-invasive tools for marine conservation This work has been accepted for publication in a Ecological Informatics and the paper can be accesed from this link.

My research is concerned with the development of novel statistical methodology for Bayesian inference and model selection for high-dimensional complex data with a particular focus on designing stochastic epidemic models and fitting them to infectious disease outbreak data.

My resume including a full list of publications and invited talks can be found here. An update to list of my publications is best found via my Google Scholar account.