Allinson, N. M. and Brown, M. T. and Johnson, M. J. (1989) {0,1}n space self-organising feature maps: extensions and hardware implementation. In: Artificial Neural Networks, 1989, 16-18 October 1989, London.

PDF 00051971.pdf - Whole Document Restricted to Repository staff only Download (302Kb) | Request a copy

Abstract

Discusses a technique for realising self-organising feature maps which exploit the properties of {0,1}n space. Working within the digital domain permits the generation of large fast networks using conventional computing machinery. Though the method exploits some of the methods of conventional N-tuple recognisers, such as WISARD, it differs in that it is an unsupervised learning process and that the output map is topologically organised. The authors concentrate on various extensions to the technique, including improved output map generation, reconstruction of corrupted input data by oversampling, and grey-scale input mapping; together with system realisation in hardware

Item Type:	Conference or Workshop Item (Paper)
Additional Information:	Discusses a technique for realising self-organising feature maps which exploit the properties of {0,1}n space. Working within the digital domain permits the generation of large fast networks using conventional computing machinery. Though the method exploits some of the methods of conventional N-tuple recognisers, such as WISARD, it differs in that it is an unsupervised learning process and that the output map is topologically organised. The authors concentrate on various extensions to the technique, including improved output map generation, reconstruction of corrupted input data by oversampling, and grey-scale input mapping; together with system realisation in hardware
Keywords:	Pattern recognition, Neural nets, Learning systems
Subjects:	G Mathematical and Computer Sciences > G920 Others in Computing Sciences
Divisions:	College of Sciences > Faculty of Science > Lincoln School of Computer Science
Depositing User:	Tammie Farley
Date Deposited:	20 Apr 2012 06:04
Last Modified:	13 Mar 2013 09:06
URI:	http://eprints.lincoln.ac.uk/id/eprint/5030

Actions (login required)

View Item