An exploration of a new paradigm for weightless RAM-based neural networks

Howells, G. and Fairhurst, M.C. and Rahman, F. (2000) An exploration of a new paradigm for weightless RAM-based neural networks. Connection Science, 12 (1). pp. 65-90. ISSN 0954-0091. (The full text of this publication is not available from this repository)

The full text of this publication is not available from this repository. (Contact us about this Publication)
Official URL
http://dx.doi.org/10.1080/095400900116203

Abstract

This paper introduces a novel networking strategy for RAM-based neurons which significantly improves the training and recognition performance of such networks whilst maintaining the generalization capabilities achieved in previous network configurations. A number of different architectures are introduced, each using the same underlying principles. Initially, features which are common to all architectures are described, illustrating the basis of the underlying paradigm. Three architectures are then introduced illustrating different techniques of employing the paradigm to meet differing performance specifications. The architectures are described in terms of the structure of the neurons they employ. Details of the various training and recognition algorithms employed by the architectures are supplied in order to present a complete description of the operation of this class of artificial neural network.

Item Type: Article
Uncontrolled keywords: weightless networks; one-shot learning; flexible architecture
Subjects: Q Science > QA Mathematics (inc Computing science) > QA 75 Electronic computers. Computer science
Divisions: Faculties > Science Technology and Medical Studies > School of Engineering and Digital Arts
Faculties > Science Technology and Medical Studies > School of Computing > Theoretical Computing Group
Depositing User: O.O. Odanye
Date Deposited: 02 Apr 2009 20:05
Last Modified: 29 May 2012 09:09
Resource URI: http://kar.kent.ac.uk/id/eprint/16243 (The current URI for this page, for reference purposes)
  • Depositors only (login required):