Computing Science and Mathematics Seminars,
Spring 2025
Unless otherwise state, seminars will take place in Room 4B96, Cottrell Building, University of Stirling from 13.00 to 14.00 on Friday afternoons, followed by informal discussions.
Spring 2025
| Date | Speaker | Title/Abstract |
|---|---|---|
| Friday 31 January |
Geoffrey Neumann, University of Southampton | Modelling large and diverse populations in marine microbial ecology - an ecological and computing challenge Photosynthetic planktonic microbes (phytoplankton) form the basis of most oceanic food webs and account for almost 50% of the global carbon sink. Computer simulations are essential in understanding the processes driving their distribution and how these may change in a changing climate. In all simulations there exist trade-offs, in particular how we effectively approximate organisms that number around 1027 individuals globally with huge diversity. Some simulations fall into the category of Individual Based Models, with which phylogenies can be tracked enabling us to fully understand the taxonomic diversity within the model, although such models are limited in population size. Population based models, in contrast allow an effectively unlimited population size but without such ability to track diversity. In this I provide a brief introduction to phytoplankton modelling and introduce a novel approach hybridising population and individual based models. I discuss, in particular, what our findings say about the role of population size and what implications this has for the future of modelling.
Speaker bio: Dr Geoffrey Neumann is a Postdoctoral Research Fellow in Modelling Marine Microbial Biodiversity at the University of Southampton. As an ecological modeller, his research focuses on uncovering global patterns in plankton biogeography. Dr Neumann has a background in computer science, he earned a BSc in 2010 and a PhD in 2014, followed by research in the same field at the University of Stirling. In 2022, he obtained an MSc in Oceanography at Southampton. Now, he combines his computer science expertise with his growing knowledge of marine ecology to design and run simulations that model populations of billions of plankton cells circulating the globe, being selected out by the environment, growing, dying, and evolving. |
| Friday 7 March |
Reading week |
|
| Friday 21 March |
Piotr Lipiński, University of Wroclaw, Poland | Using hidden data structures in explanation and optimization Modern learning algorithms often construct internal data representations in an auxiliary vector space, called embeddings, that preserve important data patterns, reflect complex relationships between the data samples and reveal some hidden data structures. This talk focuses on using additional knowledge encoded in embeddings in real-world applications, especially related to explanation and optimization. I will discuss modeling relations between frequent patterns in stock market data, discovering seasonalities in forecasting water demand in urban water distribution networks, as well as, constructing explanation rules for session-based recommender systems. Recent representation learning approaches are efficient for solving tasks with complex data structures: multivariate time series processing (e.g. Time2Vec, TRep), sequential or session-based recommender systems (e.g. LightGCN, SRGNN, TAGNN), remote sensing time series (e.g. Presto), etc. Studying such hidden data structures may lead to additional knowledge, not only about the operation of the learning algorithm, but above all about the characteristics of the data themselves, useful in explanation and optimization.
Speaker bio: Piotr Lipiński is an associate professor at the University of Wroclaw (Poland) and the head of the Computational Intelligence Research Group at the Institute of Computer Science. He received a PhD degree from the University of Strasbourg (France) in 2004, a DSc degree (habilitation) from the University of Wroclaw (Poland) in 2018, where he became an associate professor in 2023. His research interests focus on computational intelligence, evolutionary algorithms, machine learning and their real-world applications. |
| Friday 28 March |
Kelechi Samuel Nwosu and Osamu Takahashi |
13:00-13:30 Talk 1: Advancing human body measurement estimation from images using deep learning for commercial applications This research explores deep learning techniques for accurate human body measurement estimation from images, enhancing applications in e-commerce, healthcare, and ergonomics. By leveraging AI and reference objects, it improves precision and commercial viability, advancing the state of the art in computer vision-based measurement Speaker bio: Nwosu Kelechi holds a Bachelor's degree in Mechanical Engineering and has five years of experience in procurement and supply chain management and a year building and running a technology start-up. He is currently completing an MSc in Artificial Intelligence at the CSM department of the University of Stirling, where he focuses on AI-driven solutions. His research interests lie at the intersection of deep learning and computer vision, particularly in applying these technologies to human body measurement estimation for commercial use cases.
13:30-14:00 Talk 2: Exponential analysis of operational deposits Recent advances in exponential analysis have led to considerable success in several signal processing applications. However, its potential impact on financial data analysis remains largely unexplored. We investigate using exponential analysis to assist banks in monitoring their minimum cash reserves, ensuring compliance with the Liquidity Coverage Ratio (LCR) introduced by Basel III in 2008. Prony-based methods decompose financial time series into structural components, identifying depositor activity cycles and their significance. The extracted parameters provide insights into future operational trends, while fluctuations in deposit volumes and spending patterns help assess depositor risk. This work is a collaboration between the University of Stirling (Annie Cuyt, Wen-shin Lee, Anthony O’Hare), and Belgium’s University of Antwerp (Annie Cuyt, Michele Pugno) and KBC Bank (Irem Yaman). Speaker bio: Osamu Takahashi is an early-stage researcher affiliated with the University of Stirling in the EU-funded EXPOWER project. His current research focuses on exploring exponential analysis for financial time series data. He holds a Master’s degree in Economic Management and Policy, with a focus on International Financial Economics, from the University of Strathclyde, and a second Master’s degree in Mathematics and Data Science from the University of Stirling. He has work experience in fund-data infrastructure administration and export-import management, where he applied his knowledge of economics and data management to real-world situations. |
| Friday 4 April |
Wiktoria Kulik, Accenture | The challenges of digital ethics & responsible AI As adoption of Generative AI solutions increases and more regulatory scrutiny is given to AI more broadly, organisations across the world are facing growing challenges in implementing responsible AI at scale. This talk will explore the key challenges organisations encounter in adopting responsible AI, including establishment and operationalisation of key principles, and navigating evolving regulations. We will examine how organisations can build effective governance frameworks that balance innovation with ethical considerations, and the role of regulators in setting clear guidelines for AI deployment. Additionally, we will discuss the challenges of fostering a culture of responsibility within organisations, particularly in fast-moving AI environments. Speaker bio: Wiktoria Kulik is a Responsible AI Manager at Accenture, where she supports clients across multiple sectors in AI governance and responsible use of technology. She previously led Digital Ethics consulting at Sopra Steria, supporting clients in financial services and other sectors in embedding digital ethics solutions in their organisations. Wiktoria is a former policy advisor at the Centre for Data Ethics and Innovation, where she supported the launch of the UK-US Privacy Enhancing Technologies Prize Challenge and conducted research to enable the creation of trustworthy Smart Data ecosystem in the UK. She was also a member of the Oversight Group that advised the DARE UK programme. She has degrees in Philosophy and Computer Science, with specialisation in Speech and Language Processing, as well as significant experience in the tech sector. |
| Friday 11 April |
Simon Power, Computing Science and Mathematics, University of Stirling | Game theory can provide a unifying theory to produce trustworthy AI Trust in AI, and regulation of AI, are both hot topics in computer science. But while there is much empirical work in these areas, there is a lack of underlying theory. This has been problematic in the AI trust literature, which has often produced disparate, conflicting and confusing results. For example, allowing an AI system to provide an explanation is often assumed from our everyday folk psychology to increase trust. Yet some studies have found that providing explanations for an AI recommendation can have no effect on whether people follow that recommendation, leading researchers to conclude that the relationship between explainability and trust is “complicated". To make progress, we need to generate theoretically grounded hypotheses, test them with rigorous experiments, and refine them into robust models and generalisations. Similarly, current work on AI regulation lacks underlying theory to generate predictions about what type of regulatory ecosystem is optimal. In both cases, I will argue that game theory, by taking account of the potential misalignment of interests between AI developers, users, and regulators, can help us address these issues.
Speaker bio: Dr Simon Powers is a Lecturer in Trustworthy Computing Systems at the University of Stirling. He holds a PhD in Computer Science (2010) from the University of Southampton, which focussed on modelling the evolution of cooperation. He subsequently worked as a Postdoctoral Researcher in the Department of Ecology and Evolution at the University of Lausanne (2011-2015), and as a Lecturer in Computing Science at Edinburgh Napier University (2016-2024). He is an Associate Editor for IEEE Technology and Society Magazine. His research lies at the interface of computer science and the social and behavioural sciences (economics, evolutionary biology and psychology). His current work uses evolutionary game theory to model and predict the factors affecting trust between people and AI systems, and to predict the effects of different regulatory regimes on the behaviour of AI developers and users.- |
| Friday 2 May |
Hamish Carr, School of Computing, University of Leeds | Representing Energy Landscapes With Computational Topology Evolutionary computation relies on the concept of the fitness landscape, but few tools are available for understanding them. The field of computational topology has been analysing and visualising physical landscapes since the 1960s, and higher-dimensional data such as scalar fields from numerical simulations since the 1990s. This talk will give an overview of the principal topological concepts for landscape analysis and visualisation, in particular the Reeb Graph, the Contour Tree, Merge Trees, Persistent Homology, the Morse-Smale Complex and the Reeb Space, give some examples of how these tools have been used to date for energy and fitness landscapes, and show how they can potentially be used for greater understanding of evolutionary algorithms.
Speaker bio: Hamish Carr is a Professor of Computer Science at the University of Leeds, having received his Ph.D. from the University of British Columbia in 2004. His research interests include computational geometry and topology, computer graphics and scientific visualization, and algorithmic development for exascale computation. |
Previous Seminar Series
1996:
Autumn
Computing Science and Mathematics
Faculty of Natural Sciences
University of Stirling
Stirling
FK9 4LA
Scotland
UK