Graph learning

The project is about studying the connection between graph neural networks and stochastic differential equations. Stochastic differential equations are ubiquitous in many fields such as finance, physics and biology. Deep learning methods often use random noise to act as a regulariser and mitigate overfitting. The same is true for neural stochastic differential equations. The PhD project will investigate whether similar benefits are transferable to graph neural networks.

EPSRC area: information and communications technologies