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.
If you would like to give a seminar to the department in future or if you need more information, please contact the seminar organiser Ken Turner (Phone 01786-467-420, Email kjt@cs.stir.ac.uk).
Prof. Ken Turner, University of Stirling
The phases of the ANISE project (Architectural Notions In Service Engineering) will be briefly explained. The ANISE approach embodies a feature calculus that can be used to structure and analyse services. The ANISE language can be used to describe generic services as well as telecommunications services. A description will be given of the Plain Old Telephone Service using ANISE, plus some simple extensions of this. An outline strategy will be given for translating ANISE descriptions to LOTOS (Language of Temporal Ordering Specification), thus providing a formal basis. It will be shown how modular ANISE descriptions of features can be defined and then merged. Potential feature interactions can be identified statically through structural overlaps. A scenario language will be introduced to express validation tests for features in a modular fashion, and a number of examples are given. Scenarios are automatically translated to LOTOS and analysed through simulation. This allows features to be validated in isolation, and dynamically in combination with other features. The design of the translation and validation tools will be discussed, showing typical results when investigating feature descriptions. It will be explained how to extend the approach for new features.
Dr. Leslie Smith, University of Stirling
Most applied neural networks use model neurons which either completely ignore time, or consider it as a sequence of integers. Applying neural networks to real-time signal processing tasks (and all sensory tasks are of this category) entails a better model for time. Leaky integrate-and-fire neurons are a simple neural model which can provide this. I will describe the model, and discuss some applications. Proper real-time operation implies a silicon implementation of these neurons, and I will describe recent work done in collaboration with Alister Hamilton and Mark Glover at the the Department of Electrical Engineering at the University of Edinburgh on the design of analogue VLSI leaky integrate-and-fire neurons.
Prof. Alan Dearle, University of Stirling
This seminar will examine the structure of and facilities provided by traditional operating systems. I will argue that this structure is inappropriate for many application systems. I will use this to motivate the design of a new operating system, known as Charm. This operating system falls into a new class of operating systems known as exo-kernels. Charm is currently being developed at the University of Stirling by Dearle, Hulse and O'lenskie. We plan to use this new operating system for database, persistent and networked applications.
Christian W. Eurich, Department of Theoretical Physics, University of Bremen
Time delays are ubiquitous in the nervous system. In some cases, e.g. the auditory system of the barn owl, the exact timing of spikes is crucial for information processing. A Hebbian learning mechanism is presented which alters time delays rather than synaptic weights in networks of spiking neurons. The algorithm yields an equalization of delay lines resulting in the synchronization of a neuron's input signals. Possible delay adaptation mechanisms in the nervous system include the modification of synaptic properties such as the density of receptors in the postsynaptic membrane, and alterations of dendritic and axonal parameters.
Dr. Muffy Calder , University of Glasgow
It has been very difficult to find formalisms for modelling and analysing telecommunications services which offer both the appropriate language constructs and powerful (semi) automated reasoning tools. In this talk I will consider how Promela (process meta language) and the model checker SPIN might meet the requirements. A network of basic call models, with a simple asynchronous communication protocol, will be developed in Promela and then analysed using a variety of techniques, including model checking for linear temporal logic formulae using XSpin. The results are a little surprising! Along the way, I will introduce the key concepts of Promela, XSpin, and model-checking.
Up one level to Computing Science Seminars