A Summary of Past Neural Networks Research

Parallel vision using neural networks

This is a PhD project that aims to investigate the construction of a parallel vision system inspired by neuroscience. The system is based on large numbers of communicating associative memories.

Staff on project: Stephen Hobson.

Easy to use Grids

This PhD project aims to develop a Grid Portal that is simple to use, allowing users to deploy software and applications on the Grid with little or no experence.

Staff on project:  Agah Ogungboye.

Development of a Renal function analysis tool.

This project aims to develop tools that will enhance the analysis of data from renal function measurement. It is funded by the white rose consortium.

Staff on project: Garry Hollier

iREAD

This project is funded by JISC and is aims to provide an in-depth evaluation of the iRODS system to investigate and demonstrate the capabilities of iRODS and to assess its impact and potential benefit to the UK e-Science community. This evaluation will address iRODS compatibility with existing e-Science infrastructure and assess its integration with leading edge security frameworks for VO management. The evaluation will be driven from real world VO requirements and Use-Cases derived from a major UK e-Science grid project, CARMEN. A reference demonstration implementation of the new iRODS system will be built, which will be accessible to the wider e-Science community. The project team have extensive experience of large-scale implementations of SRB systems and of the requirements for VO’s within active Grid environments. This knowledge will be brought to bear in building a reference iRODS system on the White Rose Grid.

Staff on project: Mark Jessop, Tom Jackson, Aaron Turner, Mark Hewitt.

FEDAURA: Finding Fraud in The Benefits Agency

This project was supported under the DTI/EPSRC MI programme and is undertaken in collaboration with Sun, EDS, SEMA and the Department of Work and Pensions (DWP). Details can be found at the FEDAURA web site.

Detecting Drift in Aircraft Strain Gauges

This project was undertaken in collaboration with British Aerospace System Ltd. The aim of the project was to develop methods for the detection of drift in strain gauges used in aircraft monitoring applications.

Staff: Gary Hollier, Paul Zannelli, Simon Hickinbotham.

EUREDIT: Imputation of Missing Data

The problem of missing data in database records can be a problem for statisticians who need to get clear inferences from the data. This project, funded under the EU 5th framework, investigated the use of AURA based methods to impute (fill in) missing data in survey data. The project was undertaken with a number of other EU organisations and companies who are investigating the use of other methods. Full details of the project can be found on the EUREDIT web site, and its relation to other work at York is discussed here.

Staff: Ken Lees.

Financial Time Series Prediction

This project used AURA based methods to perform fast future predictions in time series data. The system also provides a 'distribution' forecast' that is a probability distribution of the likely future events.

Staff on project: Andy Pasley, Dan Kustrin.

High Performance Scheduling

This project investigatiedthe use of neural networks and other methods to perform scheduling tasks. The problem has been reduced to one of solving NP hard problems in a reasonable amount of time.

Staff on project: Serge Massicotte.

Distributed Aircraft Management Environment (DAME)

Follwed by BROADEN. See the DAME web site

BROADEN

Follow on from Distributed Aircraft Maintenance Environment (DAME). See the DAME/BROADEN web site

CATS-I

This project aimed to combine novel colour invariance methods with high performance pattern recognition methods based on our trade mark database system. The project was a collaboration with Graham Finlayson at the University of East Anglia, Ned Graphics Ltd. and David Evans Ltd.

Staff on project: Sujeewa Alwis.