DEPARTMENT OF COMPUTING SCIENCE AND MATHEMATICS
UNIVERSITY . COMPUTING SCIENCE . SEMINARS

SEMINARS - SPRING 2003

[Talk Schedule] [Abstracts] [Previous Seminars]

The Department of Computing Science and Mathematics presents the following seminars. Unless otherwise stated, seminars will take place in Room 4B94 of the Cottrell Building, University of Stirling from 15.00 to 16.00 on Friday afternoons during semester time.

If you would like to give a seminar to the department in future or if you need more information, please contact the seminar organisers, either David Cairns (Phone 01786 467445, Email dec@cs.stir.ac.uk) or Julie Hodgkin (Phone 01786 467446, Email jho@cs.stir.ac.uk).

Talk Schedule [Top] [Abstracts]

Abstracts [Top] [Schedule]

21st February 2003 [Schedule]

Agent Trees: A new internet search tool
Martin Gill, Department of Computing Science and Mathematics, University of Stirling

Information on the internet is growing at an astounding rate. However, much of the information cannot be indexed by the search engines. This is partly due to the amount of information, but also because much of the information is stored in databases and accessed via middleware products such as Asp and Cold Fusion.

During the presentation I will describe the different types of internet based systems; comparing them and giving the advantages and disadvantages that they offer for information processing and retrieval. The main part of the seminar will detail the Agent Trees system. This is a distributed multi-agent system which is intended to address some of the problems of the existing systems.

 

28th February 2003 [Schedule]

Modelling in Communications - A Multi-perspective View
Iain Buchanan, Ariadne Analytics Ltd, Emeritus Professor of Computer Science, University of Strathclyde

Based on a career as an academic, industry-based research fellow, consultant and software developer, I will describe some models and experiences from working with the telecommunications industry. These will range from the contemporary - the OFTEL and now Competition Commission review of aspects of mobile operator charging - to at least one example of areas of research and business application. In addition to the "pure" modelling material, I will try to reflect on the implication of these experiences for teaching and research.

 

7th March 2003 [Schedule]

Reliability in an unreliable world - the story of a giant synapse in the mammalian auditory system
Dr Bruce Graham, Department of Computing Science and Mathematics, University of Stirling

A major feature of the signal transmission system between neurons in our brains is that it is inherently unreliable. In the cortex this unreliability is likely to be fundamental to how information is processed. In other areas, however, reliability is required and evolution has produced some remarkable structures to produce it. In this talk I will look at an example of a reliable synapse - the calyx of Held in the mammalian auditory system. This synapse is part of neuronal circuitry involved in sound source localization through the calculation of interaural time and amplitude differences. The accurate transmission and timing of signals is crucial in this system. I will describe mathematical models and computer simulations of this synapse that I have been developing in collaboration with Prof. Ian Forsythe (University of Leicester). These models are aimed both at understanding specific experimental results from Prof. Forsythe's laboratory and providing a picture of the functionality of this synapse in sound source localization.

 

14th March 2003 [Schedule]

Marbles, Darts and Hoopla or What I Got Up To During My Sabbatical
Dr. Sharon Curtis, Department of Computing Science and Mathematics, University of Stirling

In the past, theories of greedy algorithms (such as matroids and greedoids) have focused on modelling the mathematical structure of datatypes operated on by greedy algorithms. A limitation of these theories is the existence of several well-known greedy algorithms that do not fit into these theories. In contrast, my work models greedy algorithms that solve optimization problems, by characterizing the relationships between the problem structure, and the components of the greedy algorithm. This has been very successful, not only modelling all known greedy algorithms that solve optimization problems, but also providing a novel classification scheme for greedy algorithms.

During this talk, I will outline some of the greedy theories, including my own, and present my classification scheme for greedy algorithms, illustrated by several examples.

As for the marbles, darts and hoopla? Well, most of this work of mine has concerned known greedy algorithms, some of which have existed for many decades. However, I have also discovered a few myself, and these particular greedy algorithms are standardly explained with analogies involving marbles, darts, and hoopla.

 

21st March 2003 [Schedule]

Summarising paths in graphs and hyperpaths in hypergraphs
Kieran Clenaghan, Department of Computer Science, University of York

The single-source shortest path computation is conceptually simple, well-known, and has an impressive range of applications. For numerically-weighted edges, the computation finds the least-cost or shortest path from the source node to each other node. However, the computation can be defined generically for a class of edge-weight types, not just numbers. Moreover, the computation may produce, for source, s and target, t, a value that may be best described as a "summary" of the collection of paths from s to t, and it may not be identified with any individual path in the collection.

The first part of the talk will review the motivation for a generic path summary computation by illustrating some applications. Some algebra for formally identifying the genericity will be discussed.

The second part of the talk will look at the natural generalisation from graphs to hypergraphs, and path summary to hyperpath summary. The generalisation is based on generalising from an edge which has single head and single tail nodes, to a hyperedge that differs by allowing multiple tail nodes. An (s,t)-path generalises to an (s,T)-hyperpath, where T is a set of nodes, and the hyperpath is a sub-hypergraph (and in special cases a tree). Hyperpaths naturally represent certain kinds of derivations that can be made from sets of rules, and are actually well-known, if not by name. We will look at applications of hypergraph summaries, and a generic algorithm for their computation.

My interest is in calculational proofs of correctness, but in this talk I aim only to show the attractiveness of the path/hyperpath computation from a more general perspective.

 

28th March 2003 [Schedule]

Interfacing neurons and electronics
Prof. Leslie Smith, Department of Computing Science and Mathematics, University of Stirling

Nervous systems and electronic systems are very different from each other. They are tenuously connected by their use of electrical potentials, but this can mislead us into thinking them more similar than is really the case.

Yet people have wanted to connect them together for a long time, starting from experiments with electro-encephalograms (EEGs), leading on to in vivo experiments with electrodes recording potentials, then more recently to patch clamping, recording (and sometimes stimulating) actual neural depolarisation, and even more recently to in vitro stimulation and measurements. What has motivated this work has been a mixture of the quest for improving medical understanding (including more recently prosthetics), for understanding neurophysiological systems, and how these differ from electronic computations systems, and for the capability to use actual neural systems to control external systems. There are many problems in recording and stimulating neural systems. These stem partly from the nature of the ionic "processing" on the neuron membrane, and partly from the differences in requirements between the living systems and the electronic systems. The former manifest themselves in difficulties in accurate measuring, and effectively (and realistically) stimulating the neural cultures, and the latter in difficulties in keeping such instrumented cultures alive for long periods. Different groups have taken different approaches, ranging from Fromherz's neural FET, to systems which lower electrodes for brief periods and then retract them, to systems which consist of multiple (permanent) patch clamps, to systems of permanent field potential electrodes. The aim is the same: longevity, and the capability to perform bidirectional communication between neuron and electronics.

I will also discuss where this technology might be leading: from the relatively mundane (but certainly useful) field of testing neuropharmacological chemicals, to replacing damaged (or even augmenting) human senses, to the possibility of a cyborg. Different cultures have different viewpoints on applications of this technology.

 

4th April 2003 [Schedule]

Meta-modelling: An approach to managing and integrating diverse biological models
Peter Saffrey, Department of Computer Science, UCL

One of the major challenges of contemporary science is to 'scale-up' our knowledge of micro-level phenomena to yield an understanding of macro-level phenomena. This challenge is particularly evident in biology where our growing knowledge of molecular and cell biology has still to be harnessed in such a way as to give a better understanding of gross physiological issues such as the behaviour of organs.

Low-level biological behaviour is now being explored using 'biological modelling', a diverse family of techniques for representing and testing our understanding of biological systems and their emergent behaviour. Scaling up these models, through integration and combination, is a huge challenge, since each model may address a different aspect of a system, use different assumptions, or be based upon an entirely separate modelling paradigm.

The integration of models requires a detailed understanding of modelling itself. Therefore, in order to understand modelling we have constructed a meta-model, which we will eventually use as a means to integrate and scale-up models. In this talk, we will discuss our progress and preliminary findings.

 

25th April 2003 [Schedule]

Computing a better face
Dr Peter Hancock , Department of Psychology, University of Stirling

The face perception group in the Department of Psychology at Stirling makes heavy use of computers. In this talk I shall describe various lines of research. A major current project is looking at ways to improve composites of faces, generated by witnesses to a crime. We have found that morphing together composites produced by different witnesses improves recognition rates and that a new 3/4 view composites system can also be helpful. We have developed a completely new kind of face generation system: EvoFIT, which uses a principal components analysis model of faces, with an interactive evolutionary front end. Morphing technology also allows us to manipulate the shape of faces, to make them look more or less male or female. I'll explain the technology behind these systems and show how our results are leading to changes in police policy.

 

2nd May 2003 [Schedule]

Teaching Assistant Portal: An e-Learning Platform
Dr Demessie Girma, Dept. of Electronic and Electrical Engineering, University of Strathclyde

Teaching Assistant Portal (TAP) is an e-Learning tool developed by Dr D Girma and has been piloted within the Department of Electronic and Electrical Engineering, University of Strathclyde, as an e-learning system as well as teaching information management tool. The pedagogical philosophy behind TAP, as the name suggests, is to assist instructors in delivering their course materials and monitor learners’ progress, rather than provide a 'canned' or 'packaged teaching materials' as commercial e-learning systems seem to emphasise. Among other things, instructors can easily create interactive quizzes, create group learning environment, publish class schedules, create electronic logbooks (for practical classes), etc, all with the convenience of 'anywhere anytime' access (to an e-learning server) with a browser. The talk will outline the system architecture in a typical departmental configuration, underlying pedagogical issues, and notable experiences over the last year or two. An abridged demonstration can be arranged.

 

9th May 2003 [Schedule]

Next Generation Measurement Science
Francisco J. Garcia, Agilent Technologies

Working in the 3G and WiFi areas, Agilent Labs recognises that to support global roaming and ubiquitous Internet access, the adoption of IPv6 will ultimately be a necessity for all mobile wireless and cellular access networks, owing chiefly to the addressing requirements of global roaming. The IPv6 protocol brings many other advantageous features such as enhanced flexibility, adaptability and programmability. These have allowed Agilent Labs to innovate new measurement techniques which have been called "Inline Measurements". Moving forward, our focus turns towards measurement science in order to develop innovative service assurance in the next generation wireless access network environment. This will include network discovery and troubleshooting tools and operations support system functionality appropriate for specific application domains such as Mobile IP and Voice-over-IP. This talk will address measurement science concepts and introduce Agilent Labs' new innovative measurement techniques and show how they might be exploited in developing service assurance products for Mobile IP.

30th May 2003 [Schedule]

New Voice Services
Prof. Ken Turner, Department of Computing Science and Mathematics, University of Stirling

Voice services were first developed for POTS (Plain Old Telephone Service) and the IN (Intelligent Network). Service creation and feature interaction have been extensively investigated for these. However these kinds of voice services are now well established, and almost passé from a research point of view. VoiceXML (Voice Extended Markup Language) has been widely adopted as the solution for IVR (Interactive Voice Response) services. It supports automated telephone enquiries, and can be linked to databases, web servers, and telephone networks. The notions of service and feature are not intrinsic to VoiceXML, and VoiceXML applications have quite different structuring mechanisms from traditional voice services. Although the talk will mainly be a tutorial on VoiceXML, it will also expose service research issues.

 

8th August 2003 [Schedule]

Agile All-Photonic Networks and Different Forms of Burst Switching
Prof. Gregor von Bochmann, School of Information Technology and Engineering, University of Ottawa
Note: Venue is Lecture Theatre A1

In the context of a Canadian research network on agile all-photonic networks (AAPN), we assume that very fast photonic space switches will be available in the not too distant future. We also assume that the large bandwidth available over a single optical wavelength can be shared among several traffic flows. This means that the network performs multiplexing in the time domain, and dynamically allocates the available bandwidth to diffrent traffic flows as the demand varies.

Our vision is a future agile all-photonic network that provides transparent photonic transmission between edge nodes residing close to the end-user. The edge nodes perform the electronic- photonic conversion, and provide for agile sharing of the photonic bandwidth among the different traffic flows.

In this talk we will give a summary of our AAPN research programme, provide some arguments for considering a very simple network architecture based on overlaid stars, and consider several modes of sharing the bandwidth of a single wavelength. We consider in particular the burst switching mode and present some new results on reducing the impact of contention losses.

Seminar Sponsor: The Vodafone Foundation

15th August 2003 [Schedule]

Quality of Service Issues for World-Wide Mobile Telephony
Prof. Gregor von Bochmann, School of Information Technology and Engineering, University of Ottawa
Note: Venue is Lecture Theatre A1

The next-generation communication infrastructures will use digital multimedia technologies and evolve largely over IP-based networks. For instance IP telephony will involve not only voice communication, but also live video and possibly shared spaces for collaboration. However, the facilities and parameters used in a particular instance of communication will depend largely on the preferences of users and the hardware/software limitations of their terminal devices.

We envision an automatic negotiation process that selects the most appropriate commu- nication parameters, which will depend on the device proles and the user preferences (user proles). A so-called user home directory could be used to store the user's quality of service and call processing preferences in a known location. Such a home directory is key to user mo- bility so that the user, possibly at some remote location, may use any device that is locally available (including mobile terminals).

We will explain how the functions of the home directory can be used for:

• the automatic selection of call quality parameters for multimedia conferencing between mobile users
• the authentication and accounting of mobile users
• providing presence information, as in certain chat facilities.

We will also discuss how the quality negotiation can be adapted to situations where a very large number of users participate in a video broadcast, and how hand-held and/or wearable devices can be integrated into this distributed application architecture, possibly leading to the distribution of some of the user profie information.

Seminar Sponsor: The Vodafone Foundation

 

Previous Seminar Series [Top] [Abstracts] [Schedule]

2002 - Autumn , Spring

2001 - Autumn , Spring

2000 - Autumn

 

Last Modified: 6th March 2003