An adaptive least squares algorithm for the efficient training of artificial neural networks

Author

Kollias, Stefanos ; Anastassiou, Dimitris

Author_Institution

Dept. of Electr. Eng., Columbia Univ., NY, USA

Volume

36

Issue

8

fYear

1989

fDate

8/1/1989 12:00:00 AM

Firstpage

1092

Lastpage

1101

Abstract

A novel learning algorithm is developed for the training of multilayer feedforward neural networks, based on a modification of the Marquardt-Levenberg least-squares optimization method. The algorithm updates the input weights of each neuron in the network in an effective parallel way. An adaptive distributed selection of the convergence rate parameter is presented, using suitable optimization strategies. The algorithm has better convergence properties than the conventional backpropagation learning technique. Its performance is illustrated, using examples from digital image halftoning and logical operations such as the XOR function

Keywords

learning systems; least squares approximations; neural nets; picture processing; Marquardt-Levenberg least-squares optimization method; XOR function; adaptive distributed selection; adaptive least squares algorithm; artificial neural networks; backpropagation learning technique; convergence rate parameter; digital image halftoning; learning algorithm; logical operations; multilayer feedforward; optimization strategies; Artificial neural networks; Backpropagation algorithms; Computer networks; Fault tolerance; Feedforward neural networks; Integrated circuit interconnections; Least squares methods; Multi-layer neural network; Neural networks; Neurons;

fLanguage

English

Journal_Title

Circuits and Systems, IEEE Transactions on

Publisher

ieee

ISSN

0098-4094

Type

jour

DOI

10.1109/31.192419

Filename

192419