Details of Grant 

EPSRC Reference:	GR/R96415/01
Title:	Robust Reliability of Neural Networks for Engineering Applications
Principal Investigator:	Worden, Professor K
Other Investigators:
Researcher Co-Investigators:
Project Partners:
Department:	Mechanical Engineering
Organisation:	University of Sheffield
Scheme:	Standard Research (Pre-FEC)
Starts:	01 February 2003	Ends:	31 January 2006	Value (£):	156,590
EPSRC Research Topic Classifications:	Control Engineering Intelligent & Expert Systems Non-linear Systems Mathematics
EPSRC Industrial Sector Classifications:
Related Grants:
Panel History:
Summary on Grant Application Form
Despite considerable research into the use of Neural Networks for various Engineering applications, the uptake of the technology by industry has been disappointing. In many cases, this can be attributed to the 'black-box' nature of neural networks which makes them resistant to traditional methods of certification and thus excludes them from safety critical applications. The proposed project is intended to establish a new appraoch to reliability assessment for nonlinear systems in general and neural networks in particular. The new approach will be centred on Ben Haim's non-probabilistic robust reliability method. The reason for moving away from probabilistic methods is that the events of interest for reliability analysis - namely failures - are rare events and are therefore associated with the tails of distributions which are usually badly characterised. The convex model or information-gap approach to the problem requires only a specification of the possible set of inputs to the system including the extreme events. The method also appea to lend itself better to the analysis of nonlinear systems like neural networks. The first stages of the work will be concerned with uncertainty propagatior through simple linear systems where analytical results will be available for establishing the benchmark credibility of the numerical algorithms developed The analysis and computation will then move through a progressively more complex sequence of system types ending with multi-input multi-output nonlinear systems, specifically neural networks. An important part of the work will be to determine normalisation procedures for the convex models.
Key Findings
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Potential use in non-academic contexts
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Impacts
Description This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk Summary Date Materialised
Sectors submitted by the Researcher
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
Project URL:
Further Information:
Organisation Website:	http://www.shef.ac.uk