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

SEMINARS - Autumn 2005

[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. For instructions on how to get to the University, please look at the following routes.

Talk Schedule [Top] [Abstracts]

Abstracts [Top] [Schedule]

16th September [Schedule]

Efficient Biological Grammar Acquisition
Chris Bryant
School of Computing, Robert Gordon University

Abstract
Linguistic methods have provided some interesting results in the recognition of complex biological signals. However the hand development of grammars is difficult and, because it requires human expertise, expensive. Thus, given the enormous volume of data arising from genome projects, there is a need to automate the acquisition of grammars from sets of biological sequences. We propose to develop an efficient method of acquiring such grammars using Inductive Logic Programming (ILP).

ILP systems are well suited to this task in the sense that biological grammars have been represented as logic programs using the Definite Clause Grammar and the String Variable Grammar formalisms. However, the speed at which ILP systems can generate biological grammars has been shown to be a bottleneck. When learning a biological grammar, training sequences must be parsed repeatedly during the search and thus the ILP system speed is dependent upon the efficiency of the parser. We tackled this speed problem by selecting Context-Free Grammar parsers, improving them with respect to biological data, and making them available to ILP systems written in Prolog. We show empirically that this reduces the overall induction time by an average amount of 60% of the time that ILP systems would normally spend parsing.

This work has been done as part of an on-going EPSRC project entitled "Efficient Biological Grammar Acquisition". The seminar will explain the background, motivation and aims of this project and then present results from the first year of the project.

23rd September [Schedule]

Cancelled
Art, Albums and AI - Applied Computational Intelligence and the INCITE Project
Kevin Swingler
Department of Computing Science and Mathematics, University of Stirling

Abstract
Kevin Swingler has recently moved from the Department of Psychology to Computing Science and Matematics to continue work on the INCITE project. The project aims to find practical and commercial application for the computational intellgence research being carried out by INCITE members. This talk describes some of those applications including a range of electronics devices used in racing cars and deep space telescopes; data mining projects in the music and motor industries; an art installation at the new Perth Concert Hall and our new project - The Electronic Sales Engine. You'll also find out what INCITE means.

30th September [Schedule]

The functionality provided by neuro-modulation in local-learning neural networks
Karla Parussel
Department of Computing Science and Mathematics, University of Stirling

Abstract

I present the conclusions drawn from the last three years of research carried out for my PhD. These show how global neuromodulators effect the dynamics of neural networks and the result this has on the behaviour of the agent. Biologically plausible neural networks, using spike timing-dependent plasticity hebbian learning, have been implemented to work as minimal disturbance systems. Comparisons are made between modulating and non-modulating networks in order to show how neuromodulators can be useful for adaptive agents.

7th October [Schedule]

Administrable Services in a Dynamic World
Dave Marples
Department of Computing Science and Mathematics, University of Stirling

Abstract
Telcordia is working to develop an administrable dynamic service environment in which service platforms supporting different capabilities can appear and disappear at will and without warning.
We believe that in a 2008 timeframe there will exist a market for enabling secure and reliable inter-working between mobile devices in many different domains. Services in mobile environments will have to work in the presence of variable local facilities - either the services themselves have to be coded to accommodate this variability, or we must provide Service Enablers to do it on a more universal basis.
This talk will report on a project supporting services that are portable across multiple devices, which are agnostic to the specific manufacturer and model of device on which they are executing and which can exploit (logically) local facilities. Most importantly, these services need to meet traditional telecom goals of availability and dependability.
Our goal is to provide a managed environment in which novel services can be deployed and to enable these services to take best advantage of the capabilities of the devices in their logical vicinity and then to 'rent' this environment out to third parties or to be responsible for its management.

14th October [Schedule]

Model Driven Architecture and Design Patterns - a practical solution?
Ali McLaren
Pivotal Integration

Abstract
One of the many challenges facing the software industry is being able to implement software solutions that will cope with the changing requirements of modern business and technology. Model Driven Architecture, MDA, from the Object Management Group is a vendor neutral approach to designing software solutions that is claimed to be able to handle this challenge. Design Patterns are recognised best practice solutions to known software design problems.

This seminar will look at the concepts behind MDA and Design Patterns and evaluate whether adoption of these practices is the ?esilver bullet?f that the software industry has been searching for. In particular will these practices stop so many projects failing before completion or failing because the implemented solution bears little resemblance to the original design specifications? Moreover, what are the incentives to businesses to adopt the best practices of MDA? Examples will be taken from the work carried out by Pivotal Integration as they attempt to design software design tools that automate the MDA design methodology.

21st October [Schedule]

Cancelled
Evolving Intelligence
John Levine
Department of Computing and Information Services, University of Strathclyde

Abstract
Designing intelligent behaviour is very hard; recognising it is much easier. Furthermore, given two behaviours we can generally recognise the more intelligent one. If a population of diverse behaviours can be evaluated, we can use artificial evolution to find behaviours which are more intelligent.

This simple and intuitive idea can be applied to a wide range of tasks which need intelligence, such as games playing, robotic control, planning, problem solving - indeed, anything which involves perceptions, decisions and action. In the talk I will present some diverse examples of evolution used to produce intelligence and explore how the evolutionary computation process should be configured for each task.

I will also speak on the limitations of the current process: it evolves agents which are specialised to the task in hand and are usually only reactive rather than reflective. Can we go beyond this and evolve true intelligence?

27th October (Thursday 2pm ) [Schedule]

Towards Self-organisation in Large Scale Dynamic Distributed Environments
Plamen L. Simeonov
Technical University Berlin

Abstract
Today's distributed computing and telecommunications become increasingly complex and require implicite organisation of information structures to keep pace with the growing heterogeneity, versatility and fidelity of the systems. Challenging research areas that require sophisticated methods for dynamic data and programm management are P2P, Grid and ad hoc mobile networking architectures.
The emerging field of autonomic systems (AS) addresses critical design issues such as dynamic spectrum management, resource sharing, bandwidth limitations and unpredictable network topology changes in an unconventional way. Its goal is the creation of computing systems that know and manage themselves. A major premise of this approach is the self-organisation of information about the system itself. However, to answer the questions of where and how data should be collected and linked together, of how they should be updated, merged and analysed to provide useful knowledge that allows the tracing, prediction and adaptation of the system behaviour in an effective, intelligent way is a non-trivial problem.
Interconnected nodes in large scale dynamic distributed environments are complex individual systems which exhibit various characteristics and functionalities. In general, these characteristics can be interpreted as resources. A primary objective of such architectures is to enable an optimal utilization of computing resources by the structured distribution and by the automatic service identification of these resources.
This talk presents HiPeer, a robust resource distribution and discovery algorithm that can be used for fast and fault-tolerant location of resources in networked P2P Grid environments. We have defined a concentric multi-ring overlay networking topology, whereon dynamic network management methods are deployed. In terms of performance, HiPeer delivers of number of lowest bounds. We demonstrate that for any De Bruijn digraph of degree d ??o 2 and diameter DDB, HiPeer constructs a highly reliable network, where each node maintains a routing table with at most 2d + 2 entries independent of the number N of nodes in the system. Further, we show that any existing resource in the network with at most d nodes can be found within at most DHiPeer = logd(N(d?|1)+d)?|1 overlay hops. This result is as close to the Moore bound as the query path length in the other best P2P proposals based on the De Bruijn digraphs. Thus, we argue that HiPeer defines a highly connected network with connectivity d and the lowest yet known lookup bound DHiPeer. Moreover, we show that any node?fs ?gjoin or leave?h operation in HiPeer implies a constant expected reorganization cost of the magnitude order of O(d) control messages.
Finally, we provide at outlook to our ongoing research in answering the questions of where and how data about system resources should be collected and linked together, of how they should be updated, merged and analysed to provide useful knowledge that allows the tracing, prediction and adaptation of the system behaviour in an effective, intelligent way.

4th November [Schedule]

Using Genetic Algorithms to Analyse X-ray Fibre Diffraction Patterns
Graeme Cameron
Department of Computing Science and Mathematics, University of Stirling

Abstract
Whilst X-ray diffraction is often associated with the study of crystals many biological tissues have significant spacial ordering and readily diffract X-rays. Extremely useful information on the molecular structure of these tissues can be determined from untreated, hydrated or even live samples, however the resultant fibre diffraction patterns are often complex with many overlapping reflections. This overlapping leads to problems when attempting to assertain the intensity which each individual reflection contributes to the overall diffraction pattern. By utilising genetic algorithms it is possible to analyse these patterns in a way which could lead to further information regarding the three dimensional structure of these tissues being uncovered.

11th November [Schedule]

A more-or-less relational database
Richard Bland
Information Services, University of Stirling

Abstract
The University's new student records system, SITS, holds its data in a relational database management system. (The production version uses Oracle, while training and demonstration versions, on standalone PCs, use other systems such as SQL Server.) But it doesn't follow all of the standard precepts for relational databases. For example, no foreign keys are declared. The seminar will look at these divergences from the academic purity of the relational model. Are they "mistakes", or are they pragmatic adaptations to the demands of the real world? The discussion will be illustrated by case studies of some of the application areas within SITS. These will include a discussion of how the system records the University's degree structure and manages students' movements within that structure.

18th November [Schedule]

End-User Driven Service Composition
Stephan Reiff-Marganiec
Department of Computing Science, University of Leicester

Abstract
The Service Oriented Architectures paradigm allows, at least in principle, for on-the-fly application generation. However, the exisiting implentation technologies -- web services -- are only supplying a part of an overall architecture that is required for automatic assembly of services by end-users.
We discuss an emerging architecture that

can support static service composition by use of flow languages,

could support fully automatic composition by using semantic web research activities and

can be used to assemble services based on user policies.

This talk presents emerging work and thus reflects on existing techologies and shows where their shortcomings are and suggests solutions for enhancement. The scope will also include a wide spectrum of views from rather pragmatic solutions to elements where formal methods are useful and reflect on some suggested approaches.

25th November [Schedule]

Handling Policy Conflicts in Call Control
Ken Turner
Department of Computing Science and Mathematics, University of Stirling

Abstract
Policies are becoming increasingly important in modern computer systems as a mechanism for end users and organisations to exhibit a level of control over software. Policies have long been established as an effective mechanism for enabling appropriate access control over resources, and for enforcing security considerations. However they are now becoming valued as a more general management mechanism for large-scale heterogeneous systems, including those exhibiting adaptive or autonomic behaviour.

In the telecommunications domain, features have been widely used to provide users with (limited) control over calls. However, features have the disadvantage that they are low-level and implementation-oriented in nature. Furthermore, apart from limited parameterisation of some features, they tend to be very inflexible. Policies, in contrast, have the potential to be much higher-level, goal-oriented, and very flexible. This talk will present joint work with Lynne Blair on an architecture and its realisation for distributed and hierarchical policies within the telecommunications domain. The work deals with the important issue of policy conflict - the analogy of feature interaction. A demonstration of the policy system will be given using the Department's SIP softswitch and SIP phones.

 2nd December [Schedule]

BPML diagram input on a tablet PC
Adrian Williamson (Graham Technology) and Alastair Donaldson (University of Glasgow)

Abstract
Adrian Williamson will present Alastair and his' work on pen based input of diagrams. A prototype solution for trying to facilitate pen based input of BPML and UML Activity diagrams, for incorporation into a commercial product is described. The approach taken is covered, and our experiences with Rubine's algorithm and hand-writing recognition engines are recounted.

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

Last Modified: 17th February 2005