About the project
Chaotic systems are abundant in both natural and man-made systems. This project is about developing new methods for predicting chaotic system dynamics, inspired by novel machine learning technologies, and is developed alongside a University consortium project attempting to model financial risk.
Chaos is abundant in both natural and man-made systems, from discrete event-driven dynamics like those found in financial markets, to continuous nonlinear dynamic systems more generally, including weather patterns and population dynamics. Chaotic behaviours, by their nature, are inherently difficult to predict due to their sensitivity. Despite the ubiquity of chaos in problem domains of interest, predicting, or even bounding, the dynamics of a chaotic system remains a significant research challenge.
In this interdisciplinary project, you will develop new methods for predicting chaotic systems, building off existing studies of different machine learning architectures, including echo state network variations and other recurrent network types. You will apply these methods to a variety of problems, from smaller well-studied systems, to more complex problems in financial risk analysis.
To support you in your project, you will be a part of the SONNETS Programme Grant, providing opportunities to work alongside experts in high-performance computing, chaos theory, machine learning, and financial modelling. You will also have access to the University of Southampton's high-performance compute resources (including the IRIDIS compute cluster, formerly in the TOP500 supercomputer list), as well as cloud compute resources.
