Advanced Computer Architectures (ACA)

The Advanced Computer Architectures Groups major aim is to undertake research into novel computer architectures. To date the Group's work has developed systems such as the Cortex II machine based on neural networks, novel search engines, new types of processors for computers and new ways to make computers recognise people.

Group Aims

The ACA group undertakes research into software and hardware for problems that involve large, complex unstructured data, with a particular emphasis on the use of Neural Networks and applications in Pattern Recognition. We work on the theory and application of novel methods as well as the development of the underlying hardware and software for the systems. The group has a strong theme in eScience and Grids. Many of our approaches are biologically inspired.

The group has strong industrial links supported by a spin off company (Cybula Ltd.) to address and commercialise problems in diverse subject areas including search large data sets & biometrics.

Senior Member: Jim Austin

Contact Person: Debbie Haverstock

Research Areas

Neural networks

Neural networks underpinning other work in several disciplines, concentrating on biologically plausible systems which can be implemented in fast, dedicated hardware
Methods for textual analysis
Methods for time series analysis
Graph matching methods for image and shape analysis.

Pattern Recognition

Image recognition for biometrics and for robot navigation and image analysis
3D face recognition.

Grid & e-Science

Grid middleware systems to allow distribution of search systems on grid computing infrastructures
e-Science systems

Hardware

Stack-based processing systems
Low power, pervasive systems
Hardware systems utilising DSPs and FPGAs to accelerate neural network methods.

Notable Work

Delivery of the Distributed Aircraft Maintenance Environment and its follow up BROADEN for grid-based analysis of aircraft engine monitoring data
Development of the White Rose Grid and associated e-Science Centre

Sponsorship is received from a variety of sources including DTI, EPSRC, HEFC, JISC, DtT, Yorkshire Forward, European Commission and industrial sponsors.

Research Projects

AURA
AURA Graphmatcher
BROADEN
CARMEN
Face recognition
Freeflow
HIPIC
PROFI
Trademark analysis
UDK
UFO
USE
VIDEOWARE
AICP
DAME
Euredit
FAR
Fedaura
AMADEUS

Industrial Collaborators

Rolls-Royce, Data Systems and Solutions, EDS Ltd.: Use of Pattern Searching and eScience methods for diagnostics.
Cybula Ltd. Groups spin out developing commercial applications of AURA.
Actor Knowledge Technologies Ltd., Use of AURA for trademark search databases.
Department for Work and Pensions, EDS Ltd. and Atos Origin, Sun Microsystems, development of improved fraud detection methods.
The Post Office, improved address matching technologies.
Hull MRI Centre, improved MRI analysis methods.
Infineon UK Ltd. Collaborator in Amadeus and HIPIC projects.
Vulcanic, MPE Ltd., Alpha-Mosaic and Broadcom, collaborators in Amadeus.
Sun Microsystems, eScience and Grids.
Evotec, GSK, Molecular search engines.

Academic Collaborators

Newcastle University
Sunderland University
Utrecht University
Free University of Berlin University
Hull University
Edinburgh University
Leeds University
Sheffield University
University of Hertfordshire
Evitech Institute Helsinki
DIBE University Genoa

Department of Computer Science