Home

About

Advanced Search

Browse by Discipline

Scientific Societies

E-print Alerts

Add E-prints

E-print Network
FAQHELPSITE MAPCONTACT US


  Advanced Search  

 
IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. I , NO. 1, JANUARY 1996 3 The Dependence Identification Neural
 

Summary: IEEE TRANSACTIONS ON NEURAL NETWORKS, VOL. I , NO. 1, JANUARY 1996 3
The Dependence Identification Neural
Network Construction Algorithm
John 0. Moody, Student Member, IEEE, and Panos J. Antsaklis, Fellow, IEEE
Abstract- An algorithm for constructing and training multi-
layer neural networks,dependenceidentification,is presentedin
this paper. Its distinctive features are that i) it transforms the
training problem into a set of quadratic optimization problems
that are solved by a number of linear equations, ii) it constructs
an appropriatenetwork to meet the training specifications,and
iii) the resultingnetwork architectureand weights can be further
refined with standardtraining algorithms,like backpropagation,
givinga significantspeedupin the developmenttime of the neural
network and decreasing the amount of trial and error usually
associated with network development.
I. INTRODUCTION
HE main tools for training multilayer feedfonvard neural
Tnetworks are gradient-basedoptimizationtechniques such
as the backpropagation (BP) algorithm developed by Rumel-
hart [191. Typical gradient descent algorithms are susceptible

  

Source: Antsaklis, Panos - Department of Electrical Engineering, University of Notre Dame

 

Collections: Engineering