Bio
Xiaoqing started her doctoral studies at The Alan Turing Institute in September 2019. She was also registered in the Department of Mathematics and Statistics at the University of Exeter.
Her PhD project primarily focused on theoretical and methodological innovation for highly multivariate and high-dimensional spatial stochastic processes, with applications to quantifying global exposure to anthropogenic pollutants. She developed a hybrid spatial graph and a mixed spatial graphical model framework, introducing novel concepts such as cross-neighbourhood and cross-Markov Random Field (cross-MRF). The integration of the framework with the cross-MRF achieves a mixed conditional approach, providing multiple genuine advancements over existing methods. Additionally, she explored the co-existence of two types of spatial stochastic processes within a unified framework, offering a potential solution to an open problem.
This project is theme-led by Professor Gavin Shaddick, externally advised by Professor James V. Zideck (FRSC, O.C.), and administratively supported by Professor Mark Kelson.
She obtained her Master’s degree in Statistics (with Distinction) from the University of Southampton, where she received the Dean's Award for Outstanding Achievement. In her Master’s thesis, she developed a Bayesian hierarchical spatial model to estimate PM2.5 concentrations across London at the aggregated local authority level. She quantified the uncertainties in these areal-aggregated estimates derived from point-level model outputs using three methods: Ordinary Monte Carlo, Markov Chain Monte Carlo, and theoretical reasoning.
Additionally, she quantified and 3-D visualised the non-compliance probabilities of exceeding the WHO’s PM2.5 compliance limit within each local authority, providing valuable insights for policymakers to make more informed decisions on environmental protection and air pollution control. This project was supervised by Professor Sujit Sahu.
During her PhD studies, she also worked on additional projects, including UK COVID-19 policy scenario simulations and Tweet analysis using Natural Language Processing (NLP) techniques.
Research interests
In general, Xiaoqing's research interests blend mathematical and statistical sciences, artificial intelligence, and natural sciences, primarily motivated by the challenges posed by ultra-complex data sets in real-world applications.
She is dedicated to developing innovative theories and methodologies that address real-world problems, with a particular focus on benefiting the public good, both in local communities and globally.
Inspired by her PhD project, she has recently developed an interest in computationally efficient inference and sampling methods, the foundational theory of deep generative models, as well as parallel algorithms, scalable stochastic optimisation, and modern learning framework.
She is also interested in statistical education and was an experienced educator before pursuing her Master's degree.
Selected publications and papers
- Chen, Xiaoqing et al. (2024). “Highly Multivariate High-dimensionality Spatial Stochastic Processes – A Mixed Conditional Approach”. arXiv:2408.10396
- Chen, Xiaoqing (2024). “Matern Correlation: A Panoramic Primer”. arXiv:2404.11427
- Chen, Xiaoqing et al. (2023). “On the Stochasticity of Reanalysis Outputs of 4D-Var”. arXiv:2304.03648.