About

Hi, I'm Robert. I am a theoretical physics PhD, now working as a machine learning researcher. My primary interests are in machine learning theory, algorithmic foundations, and mathematically grounded or methodological approaches to machine learning for scientific discovery. Currently, as I move deeper into the field, I am perhaps most motivated by the question: how should we learn, infer, simulate, represent, and make decisions about complex scientific systems when data are limited, experiments are expensive, uncertainty matters, and mechanistic structure is available?

The Stochastic Ledger is a machine learning research blog. It serves as a place for me to dump my thoughts, notes, and whatever else I might produce at one point or another. One will also find lots of discussion on quantitative modelling and advanced mathematical methods, with a strong emphasis on rigorous approaches to complex scientific and real-world problems.

Research topics

My research interests are increasingly centred on the intersections between generative modelling, active learning, reinforcement learning, and sequential decision-making, as well as inverse problems and simulation-based inference, scientific machine learning (SciML), and representation learning for scientific data. I am also interested in emerging areas such as probabilistic numerical methods, especially where they connect numerical computation with uncertainty quantification, inference, and learning.

My current research focuses primarily on applications in computational biology and physics, although I also work more broadly on quantitative modelling and general applied research problems.