White Rose University Consortium logo
University of Leeds logo University of Sheffield logo York University logo

A high performance k-NN approach using binary neural networks

Hodge, V J, Lees, K J and Austin, J L (2004) A high performance k-NN approach using binary neural networks. Neural Networks. pp. 441-458. ISSN 0893-6080

Full text available as:
[img]
Preview
Text (hodgevj5.pdf)
hodgevj5.pdf

Download (413Kb)

Abstract

This paper evaluates a novel k-nearest neighbour (k-NN) classifier built from binary neural networks. The binary neural approach uses robust encoding to map standard ordinal, categorical and numeric data sets onto a binary neural network. The binary neural network uses high speed pattern matching to recall a candidate set of matching records, which are then processed by a conventional k-NN approach to determine the k-best matches. We compare various configurations of the binary approach to a conventional approach for memory overheads, training speed, retrieval speed and retrieval accuracy. We demonstrate the superior performance with respect to speed and memory requirements of the binary approach compared to the standard approach and we pinpoint the optimal configurations. (C) 2003 Elsevier Ltd. All rights reserved.

Item Type: Article
Copyright, Publisher and Additional Information: Copyright © 2003 Elsevier Ltd. This is an author produced version of a paper published in Neural Networks. This paper has been peer-reviewed but does not include the final publisher proof-corrections or journal pagination.
Keywords: binary neural network, associative memory, correlation matrix memory, k-nearest neighbour, euclidean distance, robust encoding, quantisation, binary mapping
Academic Units: The University of York > Computer Science (York)
Depositing User: Sherpa Assistant
Date Deposited: 08 Nov 2005
Last Modified: 17 Oct 2013 14:27
Published Version: http://dx.doi.org/10.1016/j.neunet.2003.11.008
Status: Published
Refereed: Yes
Related URLs:
URI: http://eprints.whiterose.ac.uk/id/eprint/768

Actions (login required)

View Item View Item