Skip to main page content - your browser does not fully support our CSS, or is text-only.

Computing Science Seminars, Autumn 2013

Welcome to the Computing Science Seminars. Unless otherwise stated, seminars will take place in Room 4108,  Cottrell Building  from 15.00 to 16.00 on Friday afternoons during semester time. 

If you would like to give a seminar, suggest a speaker, or need more information,  please contact the seminar organiser: 

Autumn 2013

Date Presenter Title/Abstract
13 Sep
Dr Giancarlo Bigi
Assistant Professor
Computer Science Department
University of Pisa, Italy
Computing equilibria via optimization: In scientific contexts the term "equilibrium" has been widely used at least in physics, chemistry, engineering and economics within different frameworks, relying on different mathematical models. For instance, it may refer to physical or mechanical structures, chemical processes, the distribution of traffic over computer and telecommunication networks or over public roads, production competition, the economical dynamics of offer and demand. 
Many of these models (variational inequalities and non-cooperative games among others) share an underlying common structure which allows to conveniently formulate them in a unique format, the so-called abstract equilibrium problem. In this talk I will introduce it through some of the above examples and I will show how mathematical optimization can be exploited to devise algorithms for solving it.
20 Sep
Prof Terry Young
Professor of Healthcare Systems
Information Systems and Computing
Brunel University
How can academics help to generate a new health sector in the economy? This talk will set out a vision of building a new Health Sector in the Economy given the increases in health spending here in the UK and abroad. We will describe the health challenges posed to providers and users that we can expect over the coming years, and how academics can respond to these needs.  We will offer a particular focus on the role of health information in this rapidly changing environment.

Note: Prof. Terry Young is keen to bridge between disciplines, cultures and organisations using his experience of over 12 years in academia and 16˝ years in industry. Since 2001, he has built a portfolio of multi-university collaborative research in healthcare products and services.
27 Sep
Dr Sandy Brownlee
Senior Research Assistant
Computing Science and Mathematics
University of Stirling 
Fitness modelling for better optimisation and decision making: Fitness models (a.k.a. approximations, surrogates or meta-models) have found a number of uses in the field of meta-heuristics. They can be used to reduce calls to costly fitness evaluations, overcome difficulties in the search space, or to better understand the problem and search process. In this presentation I will give an overview of the concept and describe a couple of example fitness models. I will then give some results showing how fitness models can decrease runtime on challenging real-world problems, and how information mined from the model can be used to inform decision making. I will conclude with some work looking at how fitness models can help us better understand how metaheuristics solve problems.

4 Oct
Prof Dave Marples
Honorary Professor
Computing Science and Mathematics
University of Stirling 

Chief Scientist
The future power grid: The electricity distribution grid faces a number of significant challenges. As we move towards an electron-centric power infrastructure the increased demands of heat pumps and electric vehicles will place hitherto unforeseen strains on an infrastructure that was largely designed in the middle of the last century and certainly didn't consider the vagaries of  distributed and intermittent power sources such as solar, wind and wave power in mind.

This lecture considers the technological solution to address these challenges by means of overlaying a control network on the legacy infrastructure to co-ordinate the activities of both sinks and sources, to ensure the network remains balanced in the face of the highly variable supply and demand that will be characteristic of our future power grids.
Friday 11 Oct
Dr Alexandra Birch
Research Fellow
School of Informatics
University of Edinburgh
Why is statistical machine translation (SMT) such a hard problem? During the last 20 years, machine translation has made enormous advances, largely thanks to adapting statistical models originally proposed for speech recognition. Today, millions of people regularly use SMT to translate web pages and emails, and it is also being widely adopted in industry as a preprocessing step for professional human translators. However, state of the art models still struggle to overcome problems such as differences in word order, complex morphology, and domain adaptation.

In this talk I will present an analysis of the effect of these issues on machine translation performance. I will also discuss ways of measuring word order performance and use this to evaluate discriminative reordering models. Finally I will discuss the difficulties that all natural language processing models encounter when used out in the real world: having to adapt to different test domains. In particular I will mention recent work with translating the output of speech recognition systems.
Friday  18 Oct

Dr Neil Urquhart
Lecturer within the Institute for Informatics and Digital Innovation
Edinburgh Napier University
Scheduling and routing in the real world: This talk makes the assertion that vehicle routing research has produced increasingly more powerful problem solvers, but has not increased the realism or complexity of typical  problem instances. The time has come of use realistic street network data to increase the relevance and challenge of our work. A particular benefit of real world street data is the ability to support vehicle emissions modelling. Thus allowing emissions to be used as an optimisation criterion. Two on-line demonstrations are presented which demonstrate the use of GIS data obtained from Open Street Map and Google Maps.  The talk will also look at some current experiences of working with industry and solving their problems.
Friday 25 Oct
Mid Semester Break
Mid Semester Break
1 Nov
Dr Alan Brown
DI Solutions Ltd.
The next generation data warehouse: Exponential data-growth coupled with a thirst for access to real-time data has changed the face of the traditional Data Warehouse. This seminar will take a close look at the evolution of the Data Warehouse in both practice and principle and describe the  technological advancements which have resulted in the modern, dynamic, real-time, high volume Data Warehouse we see today. The journey from the simple batch-populated relational database in the 80's to the Massively Parallel Processing (MPP) distributed databases which are  dimensionally modelled, virtualized and fed by real-time data sources is an exciting one. However, it's a difficult journey that continues in the face of great challenges - challenges which are yet to be addressed.
8 Nov
No seminar
No Seminar
15 Nov
Dr Jingpeng Li
Computing Science and Mathematics
University of Stirling
Evolutionary based constructive search: theoretic and practical point of views: This talk presents my recently-proposed search framework (namely “evolutionary based constructive search”), and introduces some ideas about studying the theoretical properties of this class of algorithms by the means of Markov chain analysis. This framework is knowledge-based as the search is jointly driven by experts’ domain knowledge and evolutionary mechanism. This framework is flexible as other meta-heuristics and hyper-heuristics can be easily incorporated into the different steps of its search. This framework is general as the simple heuristic, the iterative constructive heuristic, the improvement heuristic, ruin and recreate principal, (evolutionary) squeaky wheel optimisation, and even the (1+1) evolutionary algorithm are all its special cases.
22 Nov
Dr Wendy Moncur
Reader in Sociodigital Interaction
School of Computing
University of Dundee
Providing adaptive health updates across the personal social network: The arrival of a new baby is usually an exciting event for parents. However, if the baby is very unwell and needs to receive medical attention, it is also a worrying time. Family and friends are likely to be concerned too, and want updates about the health of the baby and the wellbeing of the parents.

We investigated what information parents were willing to share, and how they adapted this information to individual members of their personal social network. Fieldwork was conducted with parents whose sick, newborn babies were being cared for in a Neonatal Unit (NNU) in a large hospital. We found that parents adapted the information they shared based on how close they were emotionally to members of their network, and to each network member’s tendency to worry and empathize. Using this understanding, we developed a prototype software tool, which created messages containing summaries of large volumes of complex medical data about the baby, plus information about the parents and the hospital. The tool automatically adapted these messages to individual members of parents’ social networks.

We believe that these findings can be generalized to the automatic adaptation of information content across the social network in other sensitive contexts, notably when users are chronically or critically ill.  (link to article)
29 Nov
2:00 PM
Prof Mark Harman
Professor of Software Engineering
Department of Computer Science
University College London
Search Based Software Engineering: This talk will explain some of the many exciting challenges that software engineering poses to the evolutionary computation and optimisation community. Software is virtual and inherently adaptive, making it better suited to evolutionary computation and optimisation than any other engineering material. As we shall see in this talk, this is leading to breakthroughs at the interface of software engineering and evolutionary computation. The talk will survey existing results in the area of research known as Search Based Software Engineering (SBSE) as well as recent developments in SBSE and Dynamic Adaptive SBSE, focussing on work at the interface of software engineering and evolutionary computation. 
Previous Seminar Series
2013:   Spring
2012:   Spring   Autumn
2011:   Spring   Autumn
2010:   Spring   Autumn
2009:   Spring   Autumn
2008:   Spring   Autumn
2007:   Spring   Autumn
2006:   Spring   Autumn
2005:   Spring   Autumn
2004:   Spring   Autumn
2003:   Spring   Autumn

Top image:
Local Optima Network of a Combinatorial Fitness Landscape. Nodes represent local optima and edges indicate that there are transitions between their attraction basins following a sequence of single mutation operations. Colors represent fitness values with red being high and blue low. Courtesy of Dr. Sebastien Verel, joint work summarised in:

G. Ochoa, S. Verel, F. Daolio and M. Tomassini (2013) Local Optima Networks: A New Model of Combinatorial Fitness Landscapes, Recent Advances in Fitness Landscapes. A. Engelbecht and H. Richter (Eds.), Springer.

Last Updated: 11 November 2013.