Sotirios Sabanis | School of Mathematics

Professor Sotirios Sabanis’ research develops mathematical algorithms at the intersection of stochastic modelling, machine learning and AI, with applications to financial data, including high-dimensional portfolio risk and AI-supported investment decision-making.

Sotirios develops cutting-edge diffusion-based optimization algorithms and generative models to advance AI in natural language processing (NLP) and computer vision, with applications in investment management and manufacturing, including object recognition.

His work appears in leading AI and machine learning journals, including the Journal of Machine Learning Research (JMLR) and Transactions on Machine Learning Research (TMLR), and has been presented at top conferences such as the International Conference on Machine Learning (ICML) and the Conference on Empirical Methods in Natural Language Processing (EMNLP).

His research develops explicit numerical algorithms for high-dimensional nonlinear stochastic systems and integrates them with data science and AI. A central focus is the design and analysis of stochastic optimizers for training neural networks, and of diffusion-based generative models, i.e. the mathematical machinery underpinning much of modern generative AI. This work rests on foundations in explicit schemes for stochastic (partial) differential equations and MCMC methods, which supply the convergence guarantees that make such algorithms provably reliable at scale.

Current AI projects

Archimedes Unit, Athena Research Centre

Sotirios is an affiliated researcher at the Archimedes Unit of the Athena Research Centre, a research hub dedicated to artificial intelligence, data science and algorithms.

Archimedes | Athena Research Centre

UK–South Korea Collaborative R&D grant (Innovate UK and KIAT)

Sotirios is Academic Principal Investigator on a joint UK–Korea project funded by Innovate UK and the Korea Institute for Advancement of Technology (KIAT), carried out with UK business partner and their partner team in South Korea. The award totals just under £1 million. The project applies new AI methods, namely diffusion-based optimisation algorithms and diffusion-based generative models, to problems in the consumer packaged-goods supply chain.

AI Research Companion (Centre for Investing Innovation)

Sotirios is Co-Investigator on the AI Research Companion projects, funded through the Centre for Investing Innovation. The work explores how existing large language models might be adapted and enhanced to synthesise the very large volumes of information behind investment decisions, drawing on public sources alongside aberdeen's proprietary data. A central aim is explainability: the tool is designed to show how its conclusions are reached, not just what they are. It is envisaged as an add-on to the capabilities of existing investment teams, with the reports and insights it generates supporting human decision-making.

Awards and fellowships

Turing Fellow, The Alan Turing Institute (2016–2024).
Director of the Centre for Investing Innovation, a £7.5 million partnership with the global asset management company aberdeen.
Co-Investigator and co-organiser of the four-week Isaac Newton Institute for Mathematical Sciences programme "Diffusions in Machine Learning: Foundations, Generative Models and Non-Convex Optimisation", hosted at The Alan Turing Institute in London.