School of Computing and Mathematical Sciences

Leslie Comrie Series 2023-24 Speaker Abstracts and Biographies

2023/2024 Seminar Programme

Wednesday 1 November 2023

Title: A data scientific perspective on number theory

Speaker: Dr Tim Oliver, University of Westminster

Abstract: Though traditionally considered “pure” mathematics, number theory has a strong tradition of experimentation. For example, in the 1790s, Legendre and Gauss, discovered what would become known as the prime number theorem through analysing tables of logarithms, and, in the 1960s, Birch and Swinnerton-Dyer formulated their eponymous conjecture after numerical exploration utilising an early computer. In this talk, I will overview implementations of supervised and unsupervised learning methodologies in the context of number theory. A highlight is the so-called “murmurations” phenomenon, which is manifest as a striking (and entirely unexpected) oscillating pattern

Wednesday 29 November 2023

Title: Deep Inverse Problems: From Plug-and-Play Methods to Implicit Neural Representations

Speaker: Dr Angelica Aviles-Rivero, University of Cambridge

Abstract: In recent years, the adoption of Plug-and-Play (PnP) methods in solving inverse problems has gained significant traction. However, recent developments in PnP algorithms have primarily focused on the integration of pretrained deep learning denoisers as priors, a process that demands substantial amounts of clean image data for denoiser pretraining. While data-driven deep learning models continue to evolve and deliver impressive results, the emergence of Implicit Neural Representations (INR) represents a noteworthy breakthrough. INR exhibits the ability to model complex and high-dimensional data without the need for explicit parameterisation. These implicit neural priors can efficiently serve as single shot denoisers within the PnP framework. The first part of this talk introduces a novel framework for single-shot PnP methods, addressing the challenges associated with data-intensive denoiser pretraining. In the second part, we delve into the implicit neural representations, introducing a function designed to harness the combined strengths of strong spatial and frequency attributes, departing from conventional methodologies. Notably, our novel technique demonstrates remarkable performance improvements across a diverse range of downstream tasks, with a particular focus on applications like CT reconstruction and denoising. Through rigorous experimentation, we provide a comprehensive understanding of the advantages offered by our approach

Wednesday 6 December 2023

Title: Machine Learning for Computer Vision

Speaker: Dr Xiaohao Cai, University of Southampton

Abstract: Computer vision seeks to automate tasks that our human visual system can do. It is an interdisciplinary field mainly focusing on gaining high-level understanding from digital images or videos. Machine/deep learning technologies have revolutionised many fields including computer vision. Their success generally relies on data quality and quantity. For the data scarcity scenarios like in medical imaging, their performance could drop significantly. Moreover, in many cases, they also lack generalisation (eg the cross-domain adaptation problem) and explanation (eg explainable AI). In this presentation, I will introduce some of our recent work in computer vision (eg segmentation and classification) targeting those challenges, such as subspace feature representations for few-shot learning, cross-domain adaptation, multilevel explainable AI, etc.

Biography: Dr Xiaohao Cai is a Lecturer (Assistant Professor equivalent) in the School of Electronics and Computer Science at the University of Southampton. He received his PhD degree in mathematics from The Chinese University of Hong Kong in 2012. He afterwards was a Postdoctoral Researcher at the Department of Mathematics of the Technische Universitat Kaiserslautern in Germany. After that he was a Research Fellow (Wellcome Trust and Isaac Newton Trust) affiliated with the Department of Plant Sciences and Department of Applied Mathematics and Theoretical Physics at the University of Cambridge. Thenceforth, before joining Southampton, he was a Research Fellow in the Mullard Space Science Laboratory (MSSL) at University College London (UCL). He is Fellow of Advance HE in the UK. He has served as a peer reviewer of over 60 international journals and has published over 50 peer reviewed papers in journals and conferences such as SIAM and IEEE transactions. He has broad multi-disciplinary research interests in applied mathematics, statistics, and computer science, with main focus and applications in image/signal/data processing, optimisation, machine learning and computer vision

Wednesday 24 January 2024

Title: The AI mathematician

Speaker: Professor Yang-Hui He, London Institute of Mathematical Sciences

Abstract: We summarize how AI can approach mathematics in three ways: theorem-proving, conjecture formulation, and language processing. Inspired by initial experiments in geometry and string theory, we present a number of recent experiments on how various standard machine-learning algorithms can help with pattern detection across disciplines ranging from algebraic geometry to representation theory, to combinatorics, and to number theory.

Wednesday 7th February 2024

Title: Computational micropolar fluid dynamics in the 21st century – Applications in biomechanical engineering simulation

Speaker: Professor O. Anwar Beg, University of Salford

Abstract: In 21st century biomechanical engineering, non-Newtonian fluid models are increasingly being deployed in a range applications. These include hemodynamics, digestive transport and bio-tribology in medical engineering, biomicrofluidics and biofuel cells. Other mechanical engineering applications include thermal ducts, nuclear reactor slurries, food processing and plastics manufacturing. A most elegant non-Newtonian model is the Eringen micropolar model (developed by the late Professor A.C. Eringen, Founder of the International Journal of Engineering Science at Princeton university in the 1960s) which generalizes the Navier-Stokes formulation to consider micro-rotation (spin) of micro-elements. This framework enables a continuum description of complex biofluid suspensions e.g. red blood cells, which is not possible with other non-Newtonian formulations. The Stokesian polar couple stress model is a very simple special case of the Eringen micropolar model. In this talk, we will describe 4 recent applications of the Eringen micropolar model in biofluid dynamics at Salford University, namely, pulsatile arterial nano-hemodynamics, metachronal (bio-inspired) ciliated propulsion, bacterial micro-organism gliding dynamics on slime and electro-kinetic peristaltic micropolar blood micro-pumps. A variety of numerical methods will be featured which are necessary for solving the complex nonlinear boundary value problems associated with these applications. These include FREEFEM++ finite element software and Mathematica symbolic software. Extensive visualization of flow characteristics will be provided. Future pathways will also be briefly described including the deployment of Eringen’s micro-stretch model (which includes axial contractions and extensions of micro-elements) and fluid-structure interaction (FSI), both of which are currently being investigated by the speaker, Professor Anwar Bég and his Multi-Physical Engineering Sciences Group (MPESG) at Salford University, UK.

Wednesday 21 February 2024

Title: Grain refinement and segregation zone modification under a pulsed electromagnetic field

Speaker: Dr Qingwei Bai, Helmholtz-Zentrum Dresden-Rossendorf Technische Universität Dresden, Germany (Currently visiting Centre for Advanced Simulations and Modelling, University of Greenwich)

Abstract: The development of the metallurgical industry was one of the outstanding achievements of the Industrial Revolution in the 18th century, making it possible to increase iron output, as well as to reduce production costs and the prices of metal products for consumers. Today, centuries later, the challenges in industrial production have shifted towards low carbon emissions, the pursuit of sustainability, and the emphasis on high quality products. As an illustration of aluminium alloy production, the energy consumption and greenhouse gas emissions of primary aluminium are 144612 MJ/tAl and 14.772 t CO2-eq/tAl, respectively, while the recycled aluminium is only 6.37% and 4.45% of the former value. However, Due to the mixture and/or accumulation of the tramp element at each remelting cycle, which have the tendency to segregate during solidification and precipitate into harmful intermetallic compounds even at small volume fractions to downgrade the material properties. Electromagnetic technology is a solution to manipulate impure elements to form finely dispersed crystal morphology and relieve segregation by improving the solute migration and temperature gradients in solidification rather than the ’naturally’ occurring coarse and brittle compounds. This process is as crucial as stirring instant coffee with a spoon, but how to utilize non-contact electromagnetic fields to control high-temperature liquid melts remains complicated because of its high dependence on solidification. In the first part of this presentation, I will introduce the industrial applications of electromagnetic fields and the challenges associated with them based on FEM models for solving Maxwell's equations. Subsequently, I will elaborate on our current solidification experiments in Diamond Light Source (UK's national synchrotron science facility) and its quantitative image processing.

Biography: Dr Qingwei Bai is a joint postdoctoral researcher at Helmholtz-Zentrum Dresden-Rossendorf (HZDR)/ Technische Universität Dresden in Germany and the University of Greenwich in UK. He received Ph.D. degree in Metallurgical Engineering as a joint doctoral candidate between SIMaP/EPM Lab/French National Centre for Scientific Research (CNRS) and Inner Mongolia University of Science & Technology in 2018. Subsequently, He served as a lecturer and researcher at Inner Mongolia University of Science & Technology in China until 2022. Dr Bai has been funded as principal investigator in 3 Chinese government projects. In 2022, he won highly globally competitive DAAD funding project in Germany. He is youth committee member of China's renewable resources industry technological innovation strategic alliance. He has served as a peer reviewer in Metallurgical and Materials Transactions B, Materials & Design, scientific reports, et al., and has published over 18 peer reviewed papers and 7 patents. His research interests revolve around magnetohydrodynamics (MHD) and metal solidification under electromagnetic fields, with a dedicated focus on applying electromagnetic fields to the metallurgical industry using FEM simulation technology. He possesses broad multidisciplinary research interests encompassing electromagnetism, metallurgy, engineering as well as quantitative image processing related to dendrite growth and solute migration.

Wednesday 6 March 2024

Title: Tropical Maths: Another New Way to Model the World

Speaker: Dr Ebrahim Patel, University of Greenwich

Abstract: Tropical mathematics is a novel tool, whose main advantage is to linearise scheduling systems that are conventionally nonlinear. To this effect, it has efficiently modelled transport timetables, manufacturing processes, as well as small scale systems, such as those in computer hardware and cellular organisms. In this talk, I will present an overview of my research and teaching experience with some of these applications, namely airport and railway scheduling. I will also present an extension of tropical mathematics that I have developed and coined as 'maxmin-omega'; I will argue that this model of dynamics on networks can prove impactful on a wider set of applications, such as disease and information spread in social networks, and generates fresh insights into network science areas such as threshold dynamics, network backbones, and the structure-dynamics interplay.

Biography: Ebrahim is a Lecturer in Mathematics and Data Science at the University of Greenwich. He first met tropical mathematics during his PhD at the University of Manchester. Most of his research has subsequently been conducted at the University of Oxford, focusing on discrete dynamical systems and network science, collaborating with engineers, computer scientists, and artists, and applying the work to industry. He is also a co-founder of The Bees, an award-winning mathematical writing group, aiming to promote the beauty of mathematics and its applications to non-experts. Previously, he was a founding faculty member of The London Interdisciplinary School.

Wednesday 20 March 2024

Title: The derivations of a quantum deformation of the first Weyl algebra

Speaker: Dr Isaac Oppong, University of Greenwich

Abstract: By a theorem of Dixmier, primitive quotients of enveloping algebras of finite-dimensional complex nilpotent Lie algebras are isomorphic to Weyl algebras. In view of this result, it is natural to consider simple quotients of positive parts of quantized enveloping algebras as quantum analogues of Weyl algebras. In this talk, we study the Lie algebra of derivations of the simple quotients of U_q^+(B2) of Gelfand-Kirillov dimension 2. For a specific family of such simple quotients, we prove that all derivations are inner (as in the case of Weyl algebras) whereas all other such algebras are quantum Generalized Weyl Algebras over a commutative Laurent polynomial algebra in one variable and have a first Hochschild cohomology group of dimension 1.

Biography: Isaac is a Lecturer in the School of Computing and Mathematical Sciences. He completed his PhD study at the University of Kent in 2022. His PhD research work investigates a quantum deformation of the second Weyl algebra: its derivation and Poisson derivations. He also holds a master’s degree in mathematical sciences from the African Institute for Mathematical Sciences and a bachelor’s degree in mathematics and economics from the University of Ghana. He has worked as a Teaching Assistant (University of Ghana, Ghana), Graduate Teaching Assistant/Tutor (University of Kent, UK), Lecturer (Colchester Institute, UK), and Postdoc Research Associate (University of Kent). His research interests lie in quantum and Poisson algebras. His current research work focuses on studying the first Hochschild cohomology group of the quantized enveloping algebras and their simple quotients (i.e., quantum Weyl algebras) and the first Poisson cohomology group of the semiclassical limits of these quantum algebras. Besides algebra, Isaac enjoys data science and is trying to interconnect his research area with data science.

Wednesday 24 April 2024

Title: A new improved model for predicting the level of chlorophyll-a in the oceans using satellite imaging

Speakers: Professor Dimitrios Stasinopoulos and Professor Robert Rigby, University of Greenwich

Abstract: The detection of the level of chlorophyll-a in the oceans using satellite imaging is a very important tool for monitoring phytoplankton abundance, eutrophication status and the risk of harmful algae blooms. A new methodology is introduced, which outperforms the current NASA methodology. This new methodology is based on Generalized Additive Models for Location, Scale, and Shape (GAMLSS). In this talk, we will introduce the GAMLSS methodology and explain its application to predicting the level of chlorophyll-a in the oceans using satellite imaging. The model uses the intensities of different colour bandwidths (obtained from the satellite imaging) to obtain a predicted distribution of the level of chlorophyll-a in each satellite pixel, producing a global map of chlorophyll-a in the oceans. Our analysis was published recently in the “Journal of Photogrammetry and Remote Sensing”: A novel algorithm for ocean chlorophyll-a concentration using MODIS Aqua data https://www.sciencedirect.com/science/article/pii/S0924271624000868 We will also discuss the appropriate usage and the advantages of GAMLSS over other statistical and machine learning techniques.

Biography: Professors Robert Rigby and Mikis Stasinopoulos are world leading experts in Distributional Regression, with extensive publications including a Read Paper to the UK Royal Statistical Society (which has over 3000 citations in google scholar) , three books, and two 4* World Leading Impact cases studies in the REF2021. They developed the GAMLSS model, which is used worldwide for distributional regression, and includes the "state of the art" method of centile estimation used by the World Health Organization amongst many others.

Tuesday 2 July 2024

Title: Tweeter Data Analysis for Emotion Prediction, Ideology Detection, Polarization and Hate and Offensive Languages

Speaker: Sanjay K Madria, Curators’ Distinguished Professor, Department of Computer Science, Missouri University of Science and Technology, Rolla, MO 65409, USA

Abstract: The adversarial impact of the Covid-19 pandemic has created a health crisis globally all over the world. This unprecedented crisis forced people to lockdown and changed almost every aspect of the regular activities of the people. Thus, the pandemic is also impacting everyone physically, mentally, and economically, and it, therefore, is paramount to analyse and understand emotional responses during the crisis affecting mental health. Negative emotional responses at fine-grained labels like anger and fear during the crisis might also lead to irreversible socio-economic damages. In this talk, I will discuss machine learning models trained using manually labelled data to detect various emotions at fine-grained labels in the Covid-19 tweets automatically. A custom Q &A RoBERTa model to extract phrases from the tweets that are primarily responsible for the corresponding emotions will also be discussed. I will also present historical emotion analysis using COVID-19 tweets. I will then further discuss deep learning models leveraging the pre-trained BERT model to detect the political ideology from the tweets for political polarization analysis. At the end, I will present some analysis on Hate and Offensive tweets by learning at fine-grain levels.

Biography: Sanjay K Madria is a Curators’ Distinguished Professor in the Department of Computer Science at the Missouri University of Science and Technology (formerly, University of Missouri-Rolla, USA). He has published over 300 Journal and conference papers in the areas of mobile and sensor computing, big data and cloud computing, data analytics and cybersecurity. He won five IEEE best papers awards in conferences such as IEEE MDM and IEEE SRDS. He is a co-author of a book (published with his two PhD graduates) on Secure Sensor Cloud published by Morgan and Claypool in Dec. 2018. He has graduated 20 PhDs and 34 MS thesis students, with 10 current PhDs. NSF, NIST, ARL, ARO, AFRL, DOE, Boeing, CDC-NIOSH, ORNL, Honeywell, and others have funded his research projects of over $25M. He has been awarded JSPS (Japanese Society for Promotion of Science) invitational visiting scientist fellowship, and ASEE (American Society of Engineering Education) fellowship. In 2012 and in 2019, he was awarded NRC Fellowship by National Academies, US. He is ACM Distinguished Scientist and served as an ACM and IEEE Distinguished Speaker He is an IEEE Senior Member as well as IEEE Golden Core Awardee.