Second European Conference on Speech Communication and Technology

Genova, Italy
September 24-26, 1991


Networks for Speech Recognition Structurally Optimised by Genetic Techniques Implemented on Parallel Hardware

Nigel Dodd (1), Donald Macfarlane (2), Chris Marland (3)

(1) Research Initiative in Pattern Recognition, Royal Signals and Radar Establishment, Malvern, UK
(2) Buckingham University, Buckingham UK
(3) The Transputer Centre, Bristol Polytechnic, Bristol, UK

Artificial Genetic algorithms have been shown to be effective in searching the space of networks. This paper describes the implementation of a family of such algorithms on various parallel computers consisting of Transputers. The benchmark problem chosen was the, so-called, ee-set taken from British Telecom's Connex database.

Full Paper

Bibliographic reference.  Dodd, Nigel / Macfarlane, Donald / Marland, Chris (1991): "Networks for speech recognition structurally optimised by genetic techniques implemented on parallel hardware", In EUROSPEECH-1991, 133-136.