About the project
Travel time in transport networks is not constant, but entails an element of variability that can be an important source of time losses. This results in uncertainty when attempting to predict it, and this is a concern for travellers, as it can cause early or, more critically, late arrival at the destination.
Time series problems arise in all areas of scientific enquiry. For example, human activity recognition from motion traces and diagnosis from medical signals such as EEG/ECG all involve data observed over time. Time series machine learning (TSML) offers the potential to contributing to a huge range of fields to give fresh insights into important applications.
TSML has specific challenges not found in traditional machine learning and requires bespoke learning algorithm to exploit the ordered nature of the data. The aims of this project are to improve existing algorithms to make them more scalable; to develop novel deep learning and ensemble approaches for the learning tasks; to improve the useability and explainability of the methods; and to apply them to case studies with our scientific and industrial collaborators such as Gt. Ormond St. Hospital.
The successful candidate will join a vibrant and active research group. We all work with the same codebase, the aeon toolkit and have a consistent track record of integrating new members. We also collaborate extensively with international partners, and there will be scope for collaborations with and visit to researchers in, for example, US, Australia, Brazil, France, Spain and Germany.
Note that allocation of PhD funding happens every one or two months, and once funded this position will close, so do not delay in applying.
