An Analysis of Hardware Configurations for an Adaptive Weightless Neural Network

Lorrentz, Pierre and Howells, Gareth and McDonald-Maier, Klaus D. (2008) An Analysis of Hardware Configurations for an Adaptive Weightless Neural Network. In: Ao, S.I. and Gelman, Len and Hukins, David W.L. and Hunter, Andrew and Korsunsky, A.M., eds. World Congress on Engineering 2008. Lecture Notes in Engineering and Computer Science , I-II. IAE, INT ASSOC ENGINEERS-IAENG, UNIT1, 1-F, 37-39 HUNG TO ROAD, KWUN TONG, HONG KONG, 00000, PEOPLES R CHINA pp. 66-71. ISBN 978-988-98671-9-5 . (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)

Abstract

This paper examines the potential offered by adaptive hardware configurations of a class of weightless neural architecture called the Enhanced Probabilistic Convergent Network targeted on a Virtex-II pro FPGA which is re configurable. The reconfiguration and adaptive capability of the Enhanced Probabilistic Convergent Network is a highly adaptive architecture offering a very fast, automated uninterrupted responses in potentially electronically harsh and isolated conditions. The hardware architecture is tested on a benchmark of unconstrained handwritten numerals from the Centre of Excellence for Document Analysis and Recognition.

Item Type: Conference or workshop item (Paper)
Additional information: World Congress on Engineering 2008 Imperial Coll London, London, ENGLAND, JUL 02-04, 2008
Uncontrolled keywords: adaptive neural network; FPGA; handwritten characters; reconfiguration
Subjects: Q Science > QA Mathematics (inc Computing science) > QA 76 Software, computer programming, > QA76.87 Neural computers, neural networks
Divisions: Faculties > Science Technology and Medical Studies > School of Engineering and Digital Arts > Image and Information Engineering
Depositing User: J. Harries
Date Deposited: 18 Apr 2009 11:39
Last Modified: 13 Jun 2014 14:15
Resource URI: http://kar.kent.ac.uk/id/eprint/14765 (The current URI for this page, for reference purposes)
  • Depositors only (login required):