A person posing for a photo.

Ben Hind

Using machine learning methods to numerically estimate the transport coefficients of quark-gluon plasma, and classify urban spaces and street art.

Ben studied a BSc in Mathematics and Physics at the Open University before graduating from Durham University with a MSc in Particles, Strings and Cosmology in 2024. His Masters thesis investigated the neural network architecture of normalizing flows to simulate equilibrium systems around the critical temperature marking a phase transition. This was specifically applied to scalar field theories and the Ising model, extending normalizing flows to discrete probability distributions. It was found that the training of the normalizing flow neural network slows down around the critical temperature.

Ben’s PhD project is multi-disciplinary, working with the Department of Mathematical Sciences and the School of Architecture. First, using machine learning methods to numerically estimate the transport coefficients of quark-gluon plasma. The goal is to train a neural network as an alternative to using Monte Carlo simulations for lattice QCD. Second, investigating the use of Convolutional Neural Networks to infer abstract qualities, such as ‘public space’, when classifying urban spaces and street art.

Before pursuing Mathematics and Physics, Ben graduated with a BA in Theology from the University of Wales in 2011. He has a continued interest in philosophical and ethical questions, particularly in relation to the training and use of large neural networks. He is looking forward to engaging with machine learning and data science in an inter-disciplinary context.