Quantum-Enhanced Learning of the Precursors to Extreme Events

About the Project

This PhD will develop quantum machine-learning methods to forecast extreme environmental hazards such as storms, floods, and wildfires. The project will explore how large environmental datasets can be compressed and learned using tensor networks combined with machine learning architectures on quantum computers.

The project will address four linked questions. First, can the high-dimensional environmental systems be compressed into low-dimensional latent representations via tensor networks that preserve and emphasise the early indicators of extreme events? Second, can quantum time-series forecasting models then learn the resulting dynamics, and detect the emergence of such indicators? Third, how can these indicators be designed to sample the time-series and output the warning to the limited output space of the quantum computer? Fourth, when forecasting rare, high-impact weather events, do quantum machine-learning models offer a real advantage over classical approaches in the presence of hardware noise?

This project will be supported by Qronon.