Journal of Automation and Information Sciences
Published 12 issues per year
ISSN Print: 1064-2315
ISSN Online: 2163-9337
SJR:
0.173
SNIP:
0.588
CiteScore™::
2
Indexed in
Examining Bp Modification in Neural Arrays of Different Dimensionalities
Volume 28,
Issue 5-6, 1996,
pp. 152-158
DOI: 10.1615/JAutomatInfScien.v28.i5-6.180
ABSTRACT
The backpropagation neural network is the most popular network architecture. A better weight update for this learning rule would be possible if we could compensate for future changes to these weights in earlier layers. To do so, we address some modifications of the backpropagation learning algorithm that use the expected value of the source. These modifications are examined in neural arrays of different dimensionalities by means of computer simulation. A method of using these networks as pattern classifiers is proposed and simulated.
326 Article views
1 Article downloads
Metrics
Articles with similar content:
Latest Issue
Modeling of Configurations Formed when Using Microneedle Systems
Properties of Large Deviations of Empirical Estimates in a Stochastic Optimization Problem for a Homogeneous Random Field
The Dynamics of One Arms Race Mathematical Model with a Delay
Some Ways to Modeling Input Data for Information Search in the Library of Standards when Solving Semantics Problems
Method for Constructing Primitive Polynomials for Cryptographic Subsystems of Dependable Automated Systems
Complete Asymptotics of Approximations by Certain Singular Integrals in Mathematical Modeling
Index, Volume 52, 2020