Ming, D. and Williamson, D. (2026) Generalized and scalable deep Gaussian process emulation. arXiv:2603.24538.
About Me
I am an Associate Professor in Mathematics and Data Analytics in the School of Management (SoM) at University College London (UCL). My research focuses on methodological and computational developments of statistical and machine learning tools to quantify uncertainties of multi-physics and multi-disciplinary decision systems. My research interests lie in the integration and Uncertainty Quantification (UQ) of cross-disciplinary models, supporting fast decision-making to tackle the ever-growing challenges in physics, environmental science, and natural hazards.
Prior to joining SoM, I did my postgraduate degrees at Imperial College London and ETH Zurich respectively. I then finished my PhD and worked as a postdoctoral researcher in the Department of Statistical Science at UCL.
I am the developer and maintainer of dgpsi, an open-source software package that provides a unified, production-ready implementation of linked and deep Gaussian process emulation and was recognised with the 2025 UCL Open Science and Scholarship Award.
News
- Try dgpsi interactively to explore its capabilities.
- I have been awarded UCL Global Engagement Funds to collaborate with Nanjing Tech University, the National University of Singapore and The Hong Kong Polytechnic University on predictive modelling and optimisation for net-zero building–EV–grid integration.
- Our paper on identifying sharp transitions in expensive simulators has been accepted for publication in SIAM/ASA Journal on Uncertainty Quantification.
- I have been promoted to Associate Professor at the UCL School of Management.
- Together with Serge Guillas and Alex Diaz, I am organising the invited session Coupling Machine Learning Models for Multi-Physics and Multi-Component Systems at COUPLED 2027 in Palma de Mallorca, Spain, in June 2027. Please reach out if you are interested in giving a talk.
- Our new preprint on scalable deep Gaussian process emulation for non-Gaussian responses is now available on arXiv.