Every year, G-Research runs a number of different PhD prizes in maths and data science with academic institutions in the UK, Europe and beyond.
Each prize is worth up to £10,000 and is open to final or penultimate year PhD students at specific universities, working across areas including machine learning, quantitative finance and mathematics.
We’re pleased to announce the next PhD prize winners, which ran in conjunction with Imperial College London
Learn more about our prizes“My research combines pure mathematics, particularly topology and geometry, with machine learning. I am interested in developing new methods for extracting notions of “shape” from data and leveraging these to better understand modern deep learning models.
“My PhD research focused on persistent homology, a tool inspired by algebraic topology that captures topological features, such as holes, in data across multiple scales. In particular, I studied duality phenomena within persistent homology, developed statistical methods for topological data analysis, and applied these techniques to investigate the generalisation properties of neural networks and the behaviour of large language model representation spaces under adversarial perturbations.
“As a postdoctoral researcher in the AIDOS (AI for Data-Oriented Science) group at the University of Fribourg, Switzerland, I continue to explore this intersection, with a particular focus on building more rigorous mathematical foundations for AI.”
“I am a PhD student in the department of mathematics at Imperial College London.
“My research focuses on theoretical aspects of sequential decision-making problems, including online learning, bandit algorithms and reinforcement learning. In particular, I explore how to optimally solve high-dimensional versions of these problems using gradient-based methods.”
“My research focuses on developing computationally efficient algorithms for probabilistic generative models and high-dimensional sampling.
“In particular, I study energy-based models, diffusion models and probabilistic inference methods, with the goal of improving the efficiency, scalability, controllability and statistical foundations of modern generative modelling.
“My recent work includes neural samplers for unnormalised distributions, efficient few-step diffusion methods, and effective test-time scaling and post-training.”
