Computing Science and Mathematics Seminars,
Spring 2023
Unless otherwise state, seminars will take place in Room 4B96, Cottrell Building, University of Stirling from 15.00 to 16.00 on Friday afternoons, followed by informal discussions.
Spring 2023
| Date | Speaker | Title/Abstract |
|---|---|---|
|
Friday 20 January |
Francisco Chiaravalloti University of São Paulo (Brazil) |
Using artificial intelligence and remote sensing for risk mapping of Aedes aegypti infestations and arbovirus occurrence: is it a practical task? This lecture aims to present themes related to the surveillance and control of mosquitoes and arboviruses, mainly Aedes aegypti mosquito, dengue, Zika, and chikungunya. We will discuss their current challenges and the necessity of new tools to identify high risk areas for mosquito and disease surveillance and control. We will try to answer the question of our title: how artificial intelligence and remote sensing could be used to identify these high-risk areas. Speaker bio: Francisco Chiaravalloti Neto is currently a Full Professor and Head of the Department of Epidemiology (2022 to 2024) at School of Public Health, University of São Paulo (USP), Brazil. He graduated in Civil Engineering (1981) from the Polytechnic School and Ph.D. (1999) from the School of Public Health, USP. He worked in the Department of Control of Endemic Diseases of the Secretary of Health of the State of São Paulo from 1983 to 2010, developing surveillance and control of Malaria, Chagas disease, arboviruses, and Leishmaniases. Dr. Chiaravalloti develops research in epidemiology of diseases whose agents are transmitted by vectors, spatial analysis in public health, and epidemiology of infectious diseases. He is currently a supervisor in the Graduate Programs in Public Health, Epidemiology, and Entomology in Public Health, at the School of Public Health, USP. He is an associate editor of Scientific Reports, BMC Infectious Diseases, and Revista Brasileira de Epidemiology. Dr. Chiaravalloti is a Scholarship in Research Productivity of National Council for Scientific and Technological Development (CNPq). |
| Friday 27 January |
. |
|
| Friday 3 February |
||
| Friday 10 February |
||
| Friday 17 February |
, | No seminar - Divisional Internal Event |
| Friday 24 February |
||
| Friday 3 March |
||
| Friday 10 March |
No seminar - Reading Week |
|
| Friday 17 March |
||
| Friday 24 March |
Wasim Ahmed Management, Work and Organisation, University of Stirling |
Researching social media platforms using pre-built tools for social network analysis Due to the increasing importance of social media platforms such as Twitter and YouTube, various tools and methodologies to study online content have been developed. As people increasingly participate in online communities for social, commercial, and civic interaction, various methods are needed to study these phenomena. This talk will provide an overview of research conducted on social media across health and sports using social network analysis. Specifically, the talk will focus on two popular software applications for social network anaysis: NodeXL and Gephi. These tools can apply various network layouts and clustering algorithms on social media data to extract rich insights. Speaker bio: Dr Wasim Ahmed is a Senior Lecturer in Digital Business at the Management School, University of Stirling. His research interests include digital business, sports, and health. His recent work has used social network analysis to analyse unstructured data from social media, particularly from Twitter and YouTube. His PhD involved completing an internship at Manchester United Football Club, working in social media analysis. |
| Friday 31 March |
Stefano Sarti University of Stirling |
Under the Hood of Transfer Learning for Deep Neuroevolution Deep-neuroevolution is the optimisation of deep neural architectures using evolutionary computation. Amongst these techniques, Fast-Deep Evolutionary Network Structured Representation (Fast-DENSER) has achieved considerable success in the development of Convolutional Neural Networks (CNNs) for image classification. In this study, variants of this algorithm are seen through the lens of Neuroevolution Trajectory Networks (NTNs), which use complex network modelling and visualisation to uncover intrinsic characteristics. We examine how evolution uses previously acquired knowledge on some datasets to inform the search for new domains with a specific focus on the architecture of CNNs. Results show that the transfer learning paradigm works as intended as networks mutate, incorporating layers from the best models trained on previous datasets. The use of specifically designed NTNs in this analysis enabled us to perceive the architectural characteristics that evolution favours in the design of CNNs. These findings provide novel insights that may inform the future creation of Deep Neural Networks (DNNs). Speaker bio: Stefano Sarti is a tutor and PhD student in the last year of his research, in the Computing department. His research interests are in the field of evolutionary machine learning. These relate to the analysis of Neuroevolution algorithms through effective visualisation techniques, to explain their inherent mechanisms that govern them and make them different from other evolutionary algorithms. His research is specifically directed to algorithms that evolve both the structure and the weights of Artificial Neural Networks. From shallow to deep networks, and the role that recombination has in neuroevolution. His interests and work also extend to discovering efficient utilisations of (indirect) plasticity encoding for said evolved Neural Networks. |
| Friday 7 April |
No seminar - Good Friday |
|
| Friday 14 April |
Burcu Can Buglalilar University of Stirling |
Deep Neural Networks for Morphologically Rich Languages Agglutinating languages are built upon words that are made up of a sequence of morphemes. Although the number of morphemes in languages like English is not many, the number of morphemes are usually more than 3 in the agglutinating languages such as Turkish, Finnish and Hungarian. It is very common to build a word having all the relevant bits related to tense, person, plural etc. Morphological segmentation of those words is one of the challenges in natural language processing especially in low-resource languages. Deep Neural Networks have been the current de-facto models used in many disciplines as well as natural language processing in the last decade. Due to the data-hungry nature of such deep learning methods, most of the studies mainly focus on resource-rich languages. There is still room to explore when it comes to resource-scarce, and particularly morphologically rich languages. In this talk, I will mainly give examples to our recent work that incorporates morphological information in such models without requiring massive amount of data. The results show that using morphological information in such models perform better than standalone models which only rely on tokens of words but not linguistically motivated units such as morphological units called morphemes. In the last few years, representation learning has been standard in natural language processing with its superior performance in almost any NLP task. With representation learning, a unit (usually a word) can be represented by a low dimensional vector which involves all the relevant features of this unit regarding syntax or semantics. Those features are learned using distributional and contextual information of words in a very large corpus. However, if the word does not exist or it is not frequent enough in the corpus, how should we represent this word in the same representation space? Most of the recent work handles this problem by processing each word as a set of characters or tokens where the representation is obtained through the word's characters. Here I will question whether a word should be represented by its characters, or its morphemes. How to represent words in agglutinating languages? Speaker bio: Burcu Can is a Lecturer in Computer Science. Her research interests mainly focus on natural language processing, computational linguistics, and on a range of machine learning models such as Bayesian learning and currently deep neural networks. Burcu Can received a Ph.D. degree in Computer Science from the University of York as a member of Artificial Intelligence Group in 2011. She worked as as an Assistant Professor at the Department of Computer Engineering, Hacettepe University, Turkey (2015-2020), which followed a Reader position at the University of Wolverhampton, UK (2020-2022). She led the Natural Language Processing Research Group (HUNLP) between 2015-2020. She was a Visiting Scholar as a member of the Artificial Intelligence Group, University of York from 2014 to 2015, and she was a Visiting Researcher at the Institute of Statistical Mathematics, Tokyo, Japan in 2019. She has been widely published in NLP journals and conferences. She served on the program committees of the major conferences and workshops in Computational Linguistics including ACL, AAAI, EMNLP, NAACL, IJCNLP, COLING; served on the workshop organizing committee of the Workshop on Representation Learning in several years in ACL. She is an associate editor on ACM Transactions on Asian and Low Resource Language Information Processing and Journal of Natural Language Engineering, and a member of the editorial board on Turkish Journal of Electrical Engineering and Computer Science. |
| Friday 16 June |
John Woodward Department of Computer Science, Loughborough University |
From No Free Lunch Theorems to Automated Design of Algorithms. Metaheuristics are a class of algorithms which sample the search space of a problem, and are successful because they only examine a tiny subset of the intractably large number of possibilities. The no free lunch theorems state that all such algorithms perform equally well over all functions. We will re-examined the intuition behind is statement. We will take a different approach and look at the case of “all programs”, not functions, and how this results in Occam’s Razor. We will look at some differences between these two cases and what different it can make to search. Finally I will make a case for the automatic design of algorithms. We will look at a couple of applications and show the practical implications of the theory. Speaker bio: John R. Woodward is Head of Dept of Computer Science at Loughborough. Previously he was a lecturer at the Queen Mary University of London and at the University of Stirling. Before that he was a lecturer for four years at the University of Nottingham. He holds a BSc in Theoretical Physics, an MSc in Cognitive Science and a PhD in Computer Science, all from the University of Birmingham. His research interests include Automated Software Engineering, particularly Search Based Software Engineering, Artificial Intelligence/Machine Learning and in particular Genetic Programming. He has over 50 publications in Computer Science, Operations Research and Engineering which include both theoretical and empirical contributions, and given over 50 talks at International Conferences and as an invited speaker at Universities. He has worked in industrial, military, educational and academic settings, and been employed by EDS, CERN and RAF and three UK Universities. |
Previous Seminar Series
Computing Science and Mathematics
Faculty of Natural Sciences
University of Stirling
Stirling
FK9 4LA
Scotland
UK