Cardiff University | Prifysgol Caerdydd ORCA
Online Research @ Cardiff 
WelshClear Cookie - decide language by browser settings

Differential evolution and sparse neural networks

Morgan, Peter Huw 2008. Differential evolution and sparse neural networks. Expert Systems -Oxford- 25 (4) , pp. 394-413. 10.1111/j.1468-0394.2008.00466.x

Full text not available from this repository.

Abstract

The aim of this work is to avoid overfitting by seeking parsimonious neural network models and hence to provide better out-of-sample predictions. The resulting sparse networks are easier to interpret as simple rules which, in turn, could give greater insight into the structure of the data. Fully connected feedforward neural networks are pruned through optimization of an estimated Schwartz model selection criterion using differential evolution to produce a sparse network. A quantity, , which indicates how close a parameter is to zero is used to estimate the number of model parameters which are being pruned out. The value of is incorporated into a function of the Schwartz information criterion to form an objective function whose maxima, as tends to zero, define parsimonious neural network models for a given data set. Since there is a multiplicity of maxima, differential evolution, with its greater capacity for global optimization, is used to optimize this objective function. The value of is progressively reduced during the evolution of the population of models in the manner of a sequential unconstrained optimization technique. The method is illustrated by results on four sets of data.

Item Type: Article
Date Type: Publication
Status: Published
Schools: Business (Including Economics)
Subjects: H Social Sciences > HA Statistics
Q Science > QA Mathematics
Q Science > QH Natural history > QH301 Biology
Q Science > QH Natural history > QH426 Genetics
Q Science > QP Physiology
Uncontrolled Keywords: Model selection ; Differential evolution ; Genetic algorithm ; Neural network
Publisher: Wiley-Blackwell
ISSN: 0266-4720
Last Modified: 04 Jun 2017 03:12
URI: http://orca.cf.ac.uk/id/eprint/18126

Citation Data

Cited 4 times in Google Scholar. View in Google Scholar

Cited 3 times in Scopus. View in Scopus. Powered By Scopus® Data

Actions (repository staff only)

Edit Item Edit Item