Wednesday 10th May 2000 (6 pm)
ABSTRACT
Many software engineering problems are fundamentally data defined. Specifications, the traditional means to define problems and the starting point for software development, are typically abstract approximations to some complex reality. But the specification-based approach to softweare engineering is hostage to poor approximation as well as specifications error, omission and inconsistency.
This talk will explain and give examples of the use of inductive computational technologies, such as neural computing and automatic rule induction, to improve classically developed software systems through both specification enhancement and direct implementation improvement (with the necessary post hoc specification changes).
The major source of examples will be derived from a current EPSRC-funded project to apply our ideas to air traffic control software in collaboration with National Air Traffic Services.
DEREK PARTRIDGE is Professor of Computer Science at the University of Exeter, UK. Prior to his appointment to the chair at Exeter in 1986, he was professor of Computer Science at New Mexico State University, USA. He returned to the UK in 1987 having spent 15 years abroad, mostly in the USA but also at the University of Nairobi, Kenya (2 years), and The University of Queensland, Brisbane, Australia (8 mths). In 1996 he was Visiting Professor at the Science University of Malaysia, Penang.
He was educated at London University: BSc Chemistry from University College in 1968, and a PhD in Computer Science from Imperial College in 1972. His research has ranged over Artificial Intelligence, Cognitive Science and Software Engineering. He is the author of many papers and a number of books on these subjects.
He is Head of Research for Computer Science within the School of Engineering and Computer Science. He has been the Principal Investigator of a number of projects that have been working towards the development of a multiversion software engineering methodology that is specifically designed for the exploitation of inductive programming technologies in software engineering and data mining. Currently, he leads an EPSRC-funded project exploring the uses of neural computing as a software technology for enhancement of air-traffic control software systems. This project involves both elucidation of the technology underlying neural computing and exploration of neural-net and decision-tree implementations of software modules as radically different (from conventionally programmed) versions in a multiversion approach to software reliability as well as a route to improving classical software specifications. Other current projects involve the prediction of medical outcomes (such as life or death in trauma cases, and risk of osteoporosis) and the prediction of likelihood of injury in basic training for army recruits.
The Evening Lectures are free to both members and non-members of SGES.
For further information contact:
Dr. Hui Liu, Department of Computer Science, Birkbeck College hui@dcs.bbk.ac.uk