Dynamically pruning output weights in an expanding multilayer perceptron neural network

Author

Amin, H. ; Curtis, K.M. ; Hayes Gill, B.R.

Author_Institution

Dept. of Electr. & Electron. Eng., Nottingham Univ., UK

Volume

2

fYear

1997

fDate

2-4 Jul 1997

Firstpage

991

Abstract

The network size for a multilayer perceptron neural network is often chosen arbitrarily for different applications, and the optimum size of the network is determined by a long process of trial and error. This paper presents a backpropagation algorithm. For a multilayer perceptron (MLP) neural network, that dynamically determines the optimum number of hidden nodes and applies a new pruning technique on output weights. A 29% reduction in the total number of output weights was observed for a handwritten character recognition problem using the new pruning algorithm

Keywords

backpropagation; multilayer perceptrons; optimisation; MLP; backpropagation; dynamically pruning output weights; expanding multilayer perceptron neural network; handwritten character recognition problem; hidden nodes; optimum network size; output weight pruning; Backpropagation algorithms; Character recognition; Intelligent networks; Mean square error methods; Monitoring; Multi-layer neural network; Multilayer perceptrons; Neural networks; Neurons; Parallel processing;

fLanguage

English

Publisher

ieee

Conference_Titel

Digital Signal Processing Proceedings, 1997. DSP 97., 1997 13th International Conference on

Conference_Location

Santorini

Print_ISBN

0-7803-4137-6

Type

conf

DOI

10.1109/ICDSP.1997.628530

Filename

628530