Bayesian inference and topological data analysis provide established machine learning tools. However, deep learning poses scalability challenges to these tools. My research focuses on Bayesian deep learning (BDL) and topological deep learning (TDL), which aim to make Bayesian and topological approaches to deep learning feasible. In the contemporary era of AI, BDL models can quantify uncertainties and assess risks, and TDL models can extract structure from big data. Healthcare is a safety-critical domain of application that interests me, as uncertainty quantification and elicitation of structure from biomedical data matter.

Interests

- Bayesian deep learning
- Topological deep learning
- Computing for healthcare

Education

PhD in statistics, 2009

University of Warwick

MSc in statistics, 2005

University of Warwick

BSc in mathematics, 2004

University of Ioannina

Professor in the mathematics of data science

Research on Bayesian deep learning and topological deep learning. Supervising undergraduate and postgraduate students. Advisor of studies of undergraduate and postgraduate students. Board member of the Data Science and Artificial Intelligence Institute.

Reader in the mathematics of data science

Conducted research on Bayesian deep learning and topological deep learning. Taught topics associated with data science. Lead of the advanced mathematics cluster of the Centre for Digital Trust and Security. Chair of the Mathematics Department forum.

Research scientist

Strategic hire in artificial intelligence. Principal investigator of two-year laboratory directed research and development (LDRD) project ‘Scalable Bayesian uncertainty quantification for neural networks’. Conducted research on Bayesian inference for neural networks.

Assistant professor in statistics

Conducted research on Markov chain Monte Carlo methodology. Taught three courses, namely ‘big data analytics’, ‘data analysis’ and ‘statistical methods’. Advisor of studies of twenty-one undergraduate students. Head of taught postgraduate programme in statistics and data analytics.

Research fellow in statistics

Conducted research on two projects. One project was related to Bayesian modelling of single-cell RNA sequencing data. The other project was related to Bayesian inference for rough differential equations.

Research associate in statistics

For 80% of work time, conducted research on Bayesian model selection via population Markov chain Monte Carlo for a biochemical pathway of Ewing sarcoma. For the remaining 20% of work time, administrated the UK-wide network on computational statistics and machine learning (NCSML).

Research associate in statistics

Conducted research on variance reduction for differential geometric Markov chain Monte Carlo methods.

Research associate in statistics

Performed statistical analysis of big genomic data sets to identify genetic determinants of blood lipid levels. Provided support for data filtering, bioinformatics and computing tasks.

Research statistician

Researched epidemiological associations between alcohol abstinence and personality disorders.

Access and search all publications.

Towards efficient MCMC sampling in Bayesian neural networks by exploiting symmetry.
In *Joint European Conference on Machine Learning and Knowledge Discovery in Databases (ECML PKDD), Research Track*.

(2023).
Approximate blocked Gibbs sampling for Bayesian neural networks.
In *Statistics and Computing*.

(2023).
(2023).
(2022).
Challenges in Markov chain Monte Carlo for Bayesian neural networks.
In *Statistical Science*.

(2022).
Distributions.jl: definition and modeling of probability distributions in the JuliaStats ecosystem.
In *Journal of Statistical Software*.

(2021).
Wide neural networks with bottlenecks are deep Gaussian processes.
In *Journal of Machine Learning Research*.

(2020).
Multiphase MCMC sampling for parameter inference in nonlinear ordinary differential equations.
In *Proceedings of the Twenty-First International Conference on Artificial Intelligence and Statistics*.

(2018).
The controlled thermodynamic integral for Bayesian model evidence evaluation.
In *Journal of the American Statistical Association*.

(2016).
RNA editing generates cellular subsets with diverse sequence within populations.
In *Nature Communications*.

(2016).